chore: add .gitignore

.claude/projects/-home-dom-ai-t2a/memory/regles_metier_t2a.md

@@ -0,0 +1,32 @@
+# Règles métier T2A — Connaissances critiques
+
+## 1. Index alphabétique CIM-10
+- Ne pas se contenter de vectoriser les codes (liste analytique)
+- Vectoriser les **index alphabétiques** : un médecin cherche "Gastrite", pas "K29.7"
+- Le lien langage naturel → code est bien plus riche dans l'index alphabétique
+
+## 2. Validité temporelle des codes CCAM
+- Chaque code CCAM a une date de début et de fin de validité
+- Si un acte est hors période de validité (supprimé ou remplacé dans une version), le **groupage plantera**
+- Le RAG doit toujours vérifier les dates de validité des codes dans les tables de référence
+- Version actuelle : CCAM V4 2025
+
+## 3. Diagnostics d'exclusion (piège IA classique)
+- Si le patient a un symptôme (R10.4 "Douleur abdominale") ET un diagnostic précis (K35.8 "Appendicite"),
+  le symptôme est **exclu** au profit du diagnostic précis
+- Règle : les codes **Chapitres I à XIV** de la CIM-10 priment sur les codes **Chapitre XVIII** (symptômes)
+- Le reranker doit implémenter cette priorisation
+
+## 4. Hiérarchie des actes CCAM (non-cumul)
+- La CCAM n'est pas que du texte, c'est de la **combinatoire**
+- Règles de non-cumul : deux actes anatomiquement incompatibles ou inclus l'un dans l'autre → **alerte**
+- Doit être vérifié selon le référentiel CCAM
+
+## 5. Sévérité CMA/CMS (nerf de la guerre GHM)
+- CMA = Complications ou Morbidités Associées
+- CMS = Complications ou Morbidités Associées Sévères
+- La détection des CMA/CMS détermine le passage du **niveau 1 au niveau 4 du GHM**
+- Différence de valorisation financière énorme
+- Le NLP doit chercher spécifiquement les **marqueurs de sévérité**
+  - Ex: "Insuffisance rénale **aiguë**" vs "**chronique**" → codes et niveaux différents
+  - Ex: "Dénutrition **sévère**" vs "modérée"

.gitignore

@@ -1,23 +1,76 @@
-# Python
-.venv/
+# === Python ===
 __pycache__/
-*.pyc
+*.py[cod]
+*.pyo
 *.egg-info/
-.pytest_cache/
-.hypothesis/
+*.egg
+dist/
+build/
+*.whl
 
-# Données générées
-output/
-input/
-data/
+# === Virtual environments ===
+.venv/
+venv/
+venv_*/
+env/
 
-# Référentiels (volumineux, non versionnés)
+# === ML Models & Data ===
+*.pt
+*.pth
+*.onnx
+*.bin
+*.safetensors
+*.h5
+*.hdf5
+*.pkl
+*.pickle
+*.npy
+*.npz
+*.faiss
+models/
+*.tar.gz
+*.zip
+
+# === Documents & Media ===
 *.pdf
-*.xls
+*.docx
 *.xlsx
+*.csv
+*.png
+*.jpg
+*.jpeg
+*.gif
+*.mp3
+*.wav
+*.mp4
 
-# Configuration locale
+# === IDE ===
+.idea/
+.vscode/
+*.swp
+*.swo
+*~
+
+# === OS ===
+.DS_Store
+Thumbs.db
+.~lock.*
+
+# === Secrets ===
 .env
+*.env
+credentials.json
+token.pickle
 
-# IDE / outils
-.claude/
+# === Logs & Cache ===
+*.log
+logs/
+.pytest_cache/
+.mypy_cache/
+.ruff_cache/
+htmlcov/
+.coverage
+
+# === Backups ===
+*_backup_*
+backups/

compare_cpam_models.py

@@ -0,0 +1,472 @@
+#!/usr/bin/env python3
+"""Comparaison qualité CPAM : multi-modèles sur 3 dossiers.
+
+Génère la contre-argumentation CPAM avec plusieurs modèles et compare :
+- Longueur et densité des arguments
+- Présence des 3 axes (médical, asymétrie, réglementaire)
+- Citations de preuves du dossier
+- Références aux sources RAG
+- Mots-clés d'asymétrie d'information
+"""
+
+import json
+import re
+import sys
+import time
+from pathlib import Path
+
+import requests
+
+STRUCTURED_DIR = Path("output/structured")
+OLLAMA_URL = "http://localhost:11434"
+MODELS = ["gemma3:12b-v2"]  # 12b avec nouveau prompt nuancé
+TIMEOUTS = {
+    "gemma3:12b": 120,
+    "gemma3:27b": 300,
+    "qwen3:14b": 180,
+    "mistral-small3.2:24b": 300,
+}
+
+# 3 dossiers variés : DP+DA, DAS long, DP court
+TEST_DOSSIERS = [
+    "183_23087212",  # DP+DA contestés
+    "228_23176885",  # DAS seul, arg long (1921c)
+    "153_23102610",  # DP seul, arg court
+]
+
+
+def load_dossier(dossier_name: str) -> dict | None:
+    dossier_dir = STRUCTURED_DIR / dossier_name
+    if not dossier_dir.exists():
+        return None
+    for f in list(dossier_dir.glob("*_fusionne_cim10.json")) + sorted(dossier_dir.glob("*_cim10.json")):
+        return json.loads(f.read_text())
+    return None
+
+
+def build_prompt(data: dict, controle: dict, sources: list[dict]) -> str:
+    """Reconstruit le prompt CPAM (identique au pipeline)."""
+    # Import du vrai builder pour garantir la cohérence
+    sys.path.insert(0, str(Path(__file__).parent))
+    from src.config import ControleCPAM, DossierMedical
+    from src.control.cpam_response import _build_cpam_prompt
+
+    dossier = DossierMedical.model_validate(data)
+    ctrl = ControleCPAM.model_validate(controle)
+    return _build_cpam_prompt(dossier, ctrl, sources)
+
+
+# Modèles incompatibles avec format:json d'Ollama (mode thinking)
+NO_FORMAT_JSON_MODELS = {"qwen3:14b", "qwen3:8b", "qwen3:32b"}
+
+
+def _parse_json_from_text(raw: str) -> dict | None:
+    """Parse du JSON depuis une réponse brute (avec ou sans markdown)."""
+    text = raw.strip()
+    # Retirer bloc markdown ```json ... ```
+    if text.startswith("```"):
+        first_nl = text.find("\n")
+        if first_nl != -1:
+            text = text[first_nl + 1:]
+        if text.rstrip().endswith("```"):
+            text = text.rstrip()[:-3]
+        text = text.strip()
+    # Essayer tel quel
+    try:
+        return json.loads(text)
+    except json.JSONDecodeError:
+        pass
+    # Trouver le premier { ... dernier }
+    brace_start = text.find("{")
+    brace_end = text.rfind("}")
+    if brace_start != -1 and brace_end > brace_start:
+        try:
+            return json.loads(text[brace_start:brace_end + 1])
+        except json.JSONDecodeError:
+            pass
+    return None
+
+
+def call_ollama(prompt: str, model: str) -> tuple[dict | None, float, str]:
+    """Appelle Ollama et retourne (parsed_json, duration_s, raw_text)."""
+    timeout = TIMEOUTS.get(model, 180)
+    use_format_json = model not in NO_FORMAT_JSON_MODELS
+
+    # Pour Qwen3 : ajouter /no_think pour désactiver le mode thinking
+    actual_prompt = prompt
+    if model in NO_FORMAT_JSON_MODELS:
+        actual_prompt = prompt + "\n/no_think"
+
+    payload = {
+        "model": model,
+        "prompt": actual_prompt,
+        "stream": False,
+        "options": {
+            "temperature": 0.1,
+            "num_predict": 4000,
+        },
+    }
+    if use_format_json:
+        payload["format"] = "json"
+
+    t0 = time.time()
+    try:
+        response = requests.post(
+            f"{OLLAMA_URL}/api/generate",
+            json=payload,
+            timeout=timeout,
+        )
+        response.raise_for_status()
+        duration = time.time() - t0
+        raw = response.json().get("response", "")
+
+        parsed = _parse_json_from_text(raw)
+        return parsed, duration, raw
+
+    except json.JSONDecodeError:
+        duration = time.time() - t0
+        return None, duration, raw
+    except Exception as e:
+        duration = time.time() - t0
+        return None, duration, str(e)
+
+
+def compute_metrics(parsed: dict | None) -> dict:
+    """Calcule les métriques de qualité."""
+    if parsed is None:
+        return {"valid_json": False}
+
+    full_text = json.dumps(parsed, ensure_ascii=False)
+
+    # 3 axes présents ?
+    has_med = bool(parsed.get("contre_arguments_medicaux"))
+    has_asym = bool(parsed.get("contre_arguments_asymetrie"))
+    has_regl = bool(parsed.get("contre_arguments_reglementaires"))
+    has_3axes = has_med and has_asym and has_regl
+
+    # Longueurs par axe
+    len_med = len(str(parsed.get("contre_arguments_medicaux", "")))
+    len_asym = len(str(parsed.get("contre_arguments_asymetrie", "")))
+    len_regl = len(str(parsed.get("contre_arguments_reglementaires", "")))
+    len_total_args = len_med + len_asym + len_regl
+
+    # Fallback ancien format
+    if not has_3axes:
+        len_total_args = max(len_total_args, len(str(parsed.get("contre_arguments", ""))))
+
+    # Preuves du dossier
+    preuves = parsed.get("preuves_dossier", [])
+    n_preuves = len(preuves) if isinstance(preuves, list) else 0
+
+    # Références structurées
+    refs = parsed.get("references", [])
+    n_refs = len(refs) if isinstance(refs, list) else 0
+
+    # Références avec citation verbatim
+    n_refs_citation = 0
+    if isinstance(refs, list):
+        for r in refs:
+            if isinstance(r, dict) and r.get("citation") and len(str(r["citation"])) > 20:
+                n_refs_citation += 1
+
+    # Mots-clés d'asymétrie
+    full_lower = full_text.lower()
+    asymetrie_kw = [
+        "biologie", "imagerie", "scanner", "irm", "échographie",
+        "traitement", "médicament", "posologie",
+        "asymétrie", "non transmis", "n'avait pas", "n'a pas eu accès",
+        "imc", "antécédent", "crp", "hémoglobine", "leucocytes",
+    ]
+    n_asymetrie = sum(1 for kw in asymetrie_kw if kw in full_lower)
+
+    # Points d'accord réels
+    accord = str(parsed.get("points_accord", ""))
+    accord_real = bool(accord) and accord.lower().strip() not in ("aucun", "aucun.", "n/a", "")
+
+    # Conclusion non vide
+    conclusion = str(parsed.get("conclusion", ""))
+    has_conclusion = len(conclusion) > 20
+
+    return {
+        "valid_json": True,
+        "has_3axes": has_3axes,
+        "len_med": len_med,
+        "len_asym": len_asym,
+        "len_regl": len_regl,
+        "len_total_args": len_total_args,
+        "n_preuves": n_preuves,
+        "n_refs": n_refs,
+        "n_refs_citation": n_refs_citation,
+        "n_asymetrie": n_asymetrie,
+        "accord_real": accord_real,
+        "has_conclusion": has_conclusion,
+        "total_len": len(full_text),
+    }
+
+
+def model_key(model: str) -> str:
+    """Clé courte pour un modèle (ex: 'gemma3:12b' → 'gemma3_12b')."""
+    return model.replace(":", "_").replace(".", "_")
+
+
+def print_multi_model(results: list[dict], models: list[str]):
+    """Affiche la comparaison multi-modèles."""
+    W = 140
+    col_w = 18
+    print("\n" + "=" * W)
+    print(f"COMPARAISON CPAM : {' vs '.join(models)}")
+    print("=" * W)
+
+    metric_labels = [
+        ("Durée (s)", "duration", True),
+        ("3 axes", "has_3axes", False),
+        ("Args médicaux", "len_med", False),
+        ("Args asymétrie", "len_asym", False),
+        ("Args réglementaires", "len_regl", False),
+        ("Total args (car.)", "len_total_args", False),
+        ("Preuves structurées", "n_preuves", False),
+        ("Références RAG", "n_refs", False),
+        ("Refs verbatim", "n_refs_citation", False),
+        ("Mots-clés asymétrie", "n_asymetrie", False),
+        ("Points d'accord", "accord_real", False),
+        ("Conclusion étayée", "has_conclusion", False),
+        ("Longueur totale", "total_len", False),
+    ]
+
+    for r in results:
+        print(f"\n{'─' * W}")
+        print(f"  {r['dossier']} / OGC {r['ogc']} — {r['titre']}")
+        print(f"  Argument CPAM : {r['arg_len']} car. | Prompt : {r['prompt_len']} car.")
+        print(f"{'─' * W}")
+
+        # Vérifier validité
+        all_valid = True
+        for m in models:
+            mk = model_key(m)
+            metrics = r.get(f"metrics_{mk}", {})
+            if not metrics.get("valid_json", False):
+                dur = r.get(f"duration_{mk}", 0)
+                print(f"  {m} : JSON INVALIDE ({dur:.1f}s)")
+                all_valid = False
+        if not all_valid:
+            continue
+
+        # Header
+        header = f"  {'Métrique':<25}"
+        for m in models:
+            short = m.split(":")[0][:6] + ":" + m.split(":")[-1] if ":" in m else m[:col_w]
+            header += f" {short:>{col_w}}"
+        print(header)
+        print(f"  {'─' * (25 + (col_w + 1) * len(models))}")
+
+        for label, key, is_duration in metric_labels:
+            row = f"  {label:<25}"
+            for m in models:
+                mk = model_key(m)
+                if is_duration:
+                    val = r.get(f"duration_{mk}", 0)
+                    row += f" {val:>{col_w - 1}.1f}s"
+                else:
+                    metrics = r.get(f"metrics_{mk}", {})
+                    val = metrics.get(key, 0)
+                    if isinstance(val, bool):
+                        row += f" {'Oui' if val else 'Non':>{col_w}}"
+                    else:
+                        row += f" {val:>{col_w}}"
+            print(row)
+
+    # Synthèse globale
+    print(f"\n{'=' * W}")
+    print("SYNTHÈSE GLOBALE")
+    print(f"{'=' * W}")
+
+    # Filtrer les résultats valides pour tous les modèles
+    valid = []
+    for r in results:
+        all_ok = all(r.get(f"metrics_{model_key(m)}", {}).get("valid_json", False) for m in models)
+        if all_ok:
+            valid.append(r)
+
+    if not valid:
+        print("  Aucun résultat valide pour tous les modèles.")
+        return
+
+    n = len(valid)
+    print(f"  Dossiers comparés : {n}")
+
+    # Header synthèse
+    header = f"\n  {'Métrique':<25}"
+    for m in models:
+        short = m.split(":")[0][:6] + ":" + m.split(":")[-1] if ":" in m else m[:col_w]
+        header += f" {short:>{col_w}}"
+    header += f" {'Meilleur':>{col_w}}"
+    print(header)
+    print(f"  {'─' * (25 + (col_w + 1) * (len(models) + 1))}")
+
+    # Durée
+    row = f"  {'Durée moy. (s)':<25}"
+    dur_vals = {}
+    for m in models:
+        mk = model_key(m)
+        avg_dur = sum(r.get(f"duration_{mk}", 0) for r in valid) / n
+        dur_vals[m] = avg_dur
+        row += f" {avg_dur:>{col_w - 1}.1f}s"
+    best = min(dur_vals, key=dur_vals.get)
+    row += f" {best:>{col_w}}"
+    print(row)
+
+    # Métriques (higher is better)
+    for label, key in [
+        ("Total args (car.)", "len_total_args"),
+        ("Preuves structurées", "n_preuves"),
+        ("Références RAG", "n_refs"),
+        ("Refs verbatim", "n_refs_citation"),
+        ("Mots-clés asymétrie", "n_asymetrie"),
+    ]:
+        row = f"  {label:<25}"
+        vals = {}
+        for m in models:
+            mk = model_key(m)
+            avg_val = sum(r.get(f"metrics_{mk}", {}).get(key, 0) for r in valid) / n
+            vals[m] = avg_val
+            row += f" {avg_val:>{col_w}.1f}"
+        best = max(vals, key=vals.get)
+        row += f" {best:>{col_w}}"
+        print(row)
+
+    # Booléens (count True)
+    for label, key in [
+        ("3 axes", "has_3axes"),
+        ("Points d'accord", "accord_real"),
+    ]:
+        row = f"  {label:<25}"
+        vals = {}
+        for m in models:
+            mk = model_key(m)
+            cnt = sum(1 for r in valid if r.get(f"metrics_{mk}", {}).get(key, False))
+            vals[m] = cnt
+            row += f" {f'{cnt}/{n}':>{col_w}}"
+        best = max(vals, key=vals.get)
+        row += f" {best:>{col_w}}"
+        print(row)
+
+    # Durées totales
+    print()
+    fastest = min(models, key=lambda m: sum(r.get(f"duration_{model_key(m)}", 0) for r in valid))
+    fastest_dur = sum(r.get(f"duration_{model_key(fastest)}", 0) for r in valid)
+    for m in models:
+        mk = model_key(m)
+        total = sum(r.get(f"duration_{mk}", 0) for r in valid)
+        ratio = total / fastest_dur if fastest_dur > 0 else 0
+        print(f"  {m:<25} total={total:.0f}s (x{ratio:.1f})")
+    print()
+
+
+def main():
+    # Charger les résultats précédents (all_models)
+    prev_file = Path("output/compare_cpam_all_models.json")
+    prev_data = {}
+    if prev_file.exists():
+        for entry in json.loads(prev_file.read_text()):
+            prev_data[entry["dossier"]] = entry
+
+    # On compare l'ancien 12b (ancien prompt) vs le nouveau 12b-v2 (nouveau prompt nuancé)
+    # + 27b comme référence nuance
+    ref_models = ["gemma3:12b", "gemma3:27b"]
+    all_models = ref_models + MODELS
+    print("=" * 100)
+    print(f"Comparaison qualité CPAM : {' / '.join(all_models)}")
+    print(f"Dossiers : {', '.join(TEST_DOSSIERS)}")
+    print(f"Test : gemma3:12b avec NOUVEAU prompt nuancé (v2)")
+    print(f"Résultats précédents : {'oui' if prev_data else 'non'}")
+    print("=" * 100)
+
+    results = []
+
+    for dossier_name in TEST_DOSSIERS:
+        data = load_dossier(dossier_name)
+        if not data:
+            print(f"\nERREUR : {dossier_name} non trouvé")
+            continue
+
+        controles = [c for c in data.get("controles_cpam", []) if c.get("arg_ucr")]
+        if not controles:
+            print(f"\nERREUR : {dossier_name} — pas de contrôle CPAM")
+            continue
+
+        controle = controles[0]
+        sources = [
+            {
+                "document": s.get("document", ""),
+                "page": s.get("page"),
+                "code": s.get("code"),
+                "extrait": s.get("extrait", ""),
+            }
+            for s in controle.get("sources_reponse", [])
+        ]
+
+        prompt = build_prompt(data, controle, sources)
+
+        print(f"\n[{dossier_name}] OGC {controle['numero_ogc']} — {controle.get('titre', '')}")
+        print(f"  Prompt : {len(prompt)} car. | Arg CPAM : {len(controle.get('arg_ucr', ''))} car.")
+
+        result = {
+            "dossier": dossier_name,
+            "ogc": controle["numero_ogc"],
+            "titre": controle.get("titre", ""),
+            "arg_len": len(controle.get("arg_ucr", "")),
+            "prompt_len": len(prompt),
+        }
+
+        # Réutiliser les résultats précédents pour les modèles de référence
+        prev = prev_data.get(dossier_name)
+        if prev:
+            for old_model in ref_models:
+                mk = model_key(old_model)
+                result[f"duration_{mk}"] = prev.get(f"duration_{mk}", 0)
+                result[f"metrics_{mk}"] = prev.get(f"metrics_{mk}", {})
+                result[f"response_{mk}"] = prev.get(f"response_{mk}")
+                dur = result[f"duration_{mk}"]
+                is_valid = result[f"metrics_{mk}"].get("valid_json", False)
+                print(f"  → {old_model} ... (précédent) {'OK' if is_valid else 'FAIL'} ({dur:.1f}s)")
+
+        # Tester le 12b-v2 (nouveau prompt) — appelle gemma3:12b avec le prompt modifié
+        for model_label in MODELS:
+            mk = model_key(model_label)
+            actual_model = "gemma3:12b"  # même modèle, nouveau prompt
+            print(f"  → {model_label} (nouveau prompt) ...", end=" ", flush=True)
+            parsed, dur, raw = call_ollama(prompt, actual_model)
+            status = "OK" if parsed else "FAIL"
+            print(f"{status} ({dur:.1f}s)")
+
+            result[f"duration_{mk}"] = dur
+            result[f"metrics_{mk}"] = compute_metrics(parsed)
+            result[f"response_{mk}"] = parsed
+
+        results.append(result)
+
+    # Affichage
+    print_multi_model(results, all_models)
+
+    # Sauvegarde
+    output_file = Path("output/compare_cpam_prompt_v2.json")
+    output_file.parent.mkdir(parents=True, exist_ok=True)
+    save_data = []
+    for r in results:
+        entry = {
+            "dossier": r["dossier"],
+            "ogc": r["ogc"],
+            "titre": r["titre"],
+        }
+        for m in all_models:
+            mk = model_key(m)
+            entry[f"duration_{mk}"] = r.get(f"duration_{mk}", 0)
+            entry[f"metrics_{mk}"] = r.get(f"metrics_{mk}", {})
+            entry[f"response_{mk}"] = r.get(f"response_{mk}")
+        save_data.append(entry)
+    output_file.write_text(json.dumps(save_data, ensure_ascii=False, indent=2))
+    print(f"Résultats sauvegardés dans {output_file}")
+
+
+if __name__ == "__main__":
+    main()

cpam/extract_t2a_llm.py

@@ -0,0 +1,738 @@
+#!/usr/bin/env python3
+"""
+extract_t2a_llm.py — Extracteur T2A généraliste via OCR + LLM (Ollama)
+
+Entrée  : PDF (scanné ou natif) de document T2A (décision UCR, notification CPAM, rapport ARS…)
+Sortie  : Fichier Excel (.xlsx) avec les données structurées
+
+Architecture :
+  PDF → OCR/texte natif → Détection type (1 appel LLM) → Extraction bloc par bloc (N appels LLM) → Excel
+
+Usage :
+  python extract_t2a_llm.py FICHIER.pdf [--model gemma3:27b-it-qat] [--output out.xlsx] [--verbose]
+"""
+from __future__ import annotations
+
+import argparse
+import json
+import re
+import sys
+import time
+from pathlib import Path
+
+import pymupdf
+import requests
+from openpyxl import Workbook
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+
+
+# ---------------------------------------------------------------------------
+# 0. Normalisation texte OCR
+# ---------------------------------------------------------------------------
+
+def normalize_text(text: str) -> str:
+    """Normalise les apostrophes, guillemets et espaces issus de l'OCR."""
+    text = text.replace("\u2018", "'").replace("\u2019", "'")
+    text = text.replace("\u201C", '"').replace("\u201D", '"')
+    text = text.replace("\u00AB", '"').replace("\u00BB", '"')
+    text = text.replace("''", "'")
+    text = text.replace("\u00A0", " ").replace("\u202F", " ")
+    text = re.sub(r"\bF'UCR\b", "l'UCR", text)
+    text = re.sub(r"\bl''UCR\b", "l'UCR", text)
+    return text
+
+
+# ---------------------------------------------------------------------------
+# 1. OCR / Extraction texte (docTR — deep learning, GPU)
+# ---------------------------------------------------------------------------
+
+_doctr_model = None
+
+
+def _get_doctr_model():
+    """Lazy-init du modèle docTR (chargé une seule fois, GPU si VRAM libre, sinon CPU)."""
+    global _doctr_model
+    if _doctr_model is not None:
+        return _doctr_model
+
+    from doctr.models import ocr_predictor
+
+    print("  Chargement du modèle docTR (première utilisation)...")
+    t0 = time.time()
+    _doctr_model = ocr_predictor(
+        det_arch="db_resnet50",
+        reco_arch="crnn_vgg16_bn",
+        pretrained=True,
+    )
+
+    # Déplacer sur GPU si disponible et assez de VRAM libre
+    try:
+        import torch
+        if torch.cuda.is_available():
+            free_vram = torch.cuda.mem_get_info()[0] / (1024 ** 3)
+            if free_vram > 1.0:
+                try:
+                    _doctr_model = _doctr_model.cuda()
+                    print(f"  docTR sur GPU ({torch.cuda.get_device_name(0)}, "
+                          f"{free_vram:.1f} Go libres) — {time.time() - t0:.1f}s")
+                except torch.cuda.OutOfMemoryError:
+                    _doctr_model = _doctr_model.cpu()
+                    torch.cuda.empty_cache()
+                    print(f"  GPU VRAM insuffisante, docTR sur CPU — {time.time() - t0:.1f}s")
+            else:
+                print(f"  GPU VRAM trop basse ({free_vram:.1f} Go libres, Ollama ?), "
+                      f"docTR sur CPU — {time.time() - t0:.1f}s")
+        else:
+            print(f"  docTR sur CPU — {time.time() - t0:.1f}s")
+    except ImportError:
+        print(f"  docTR sur CPU — {time.time() - t0:.1f}s")
+
+    return _doctr_model
+
+
+def ocr_pdf(pdf_path: str, dpi: int = 300) -> str:
+    """Extrait le texte du PDF : texte natif si disponible, sinon OCR docTR (GPU)."""
+    doc = pymupdf.open(pdf_path)
+    total = len(doc)
+
+    # Détection : texte natif vs scanné (sur la première page)
+    first_page_text = doc[0].get_text() if total > 0 else ""
+    is_native = len(first_page_text.strip()) > 100
+
+    if is_native:
+        print("  Mode : extraction texte natif (pymupdf)")
+        full_text = []
+        for i, page in enumerate(doc):
+            print(f"  Extraction page {i+1}/{total}...", end="\r")
+            full_text.append(page.get_text())
+        print(f"  Extraction terminée : {total} pages.          ")
+        return normalize_text("\n\n".join(full_text))
+
+    # OCR docTR
+    print("  Mode : OCR docTR (deep learning, GPU)")
+    from doctr.io import DocumentFile
+
+    model = _get_doctr_model()
+
+    print(f"  Lecture du PDF ({total} pages)...")
+    doc_pages = DocumentFile.from_pdf(pdf_path)
+    print(f"  OCR en cours sur {len(doc_pages)} pages...")
+
+    t0 = time.time()
+    result = model(doc_pages)
+    elapsed = time.time() - t0
+    print(f"  OCR terminé : {total} pages en {elapsed:.1f}s "
+          f"({elapsed/total:.1f}s/page)")
+
+    full_text = result.render()
+    return normalize_text(full_text)
+
+
+# ---------------------------------------------------------------------------
+# 2. Client Ollama
+# ---------------------------------------------------------------------------
+
+NO_FORMAT_JSON_PREFIXES = ("qwen3", "qwen2.5")
+
+OLLAMA_URL = "http://localhost:11434"
+
+
+def parse_json_response(raw: str) -> dict | list | None:
+    """Parse une réponse JSON, en gérant les blocs markdown et le texte parasite."""
+    text = raw.strip()
+
+    # Supprimer les blocs <think>...</think> (Qwen3)
+    text = re.sub(r"<think>.*?</think>", "", text, flags=re.DOTALL).strip()
+
+    # Supprimer les blocs markdown ```json ... ```
+    if text.startswith("```"):
+        first_nl = text.find("\n")
+        if first_nl != -1:
+            text = text[first_nl + 1:]
+        if text.rstrip().endswith("```"):
+            text = text.rstrip()[:-3]
+        text = text.strip()
+
+    # Tentative directe
+    try:
+        return json.loads(text)
+    except json.JSONDecodeError:
+        pass
+
+    # Extraire le premier objet ou tableau JSON
+    for start_char, end_char in [("{", "}"), ("[", "]")]:
+        start = text.find(start_char)
+        if start == -1:
+            continue
+        depth = 0
+        for i in range(start, len(text)):
+            if text[i] == start_char:
+                depth += 1
+            elif text[i] == end_char:
+                depth -= 1
+            if depth == 0:
+                try:
+                    return json.loads(text[start:i + 1])
+                except json.JSONDecodeError:
+                    break
+
+    return None
+
+
+def call_ollama(
+    prompt: str,
+    model: str,
+    temperature: float = 0.1,
+    max_tokens: int = 4000,
+    timeout: int = 120,
+    verbose: bool = False,
+) -> dict | list | None:
+    """Appelle Ollama. Utilise l'API chat avec think=false pour Qwen3."""
+    is_qwen = any(model.startswith(p) for p in NO_FORMAT_JSON_PREFIXES)
+
+    if is_qwen:
+        # API chat + think:false pour Qwen3 (pas de format JSON natif)
+        endpoint = f"{OLLAMA_URL}/api/chat"
+        body = {
+            "model": model,
+            "messages": [{"role": "user", "content": prompt}],
+            "stream": False,
+            "think": False,
+            "options": {
+                "temperature": temperature,
+                "num_predict": max_tokens,
+            },
+        }
+    else:
+        # API generate + format JSON natif pour les autres modèles
+        endpoint = f"{OLLAMA_URL}/api/generate"
+        body = {
+            "model": model,
+            "prompt": prompt,
+            "stream": False,
+            "format": "json",
+            "options": {
+                "temperature": temperature,
+                "num_predict": max_tokens,
+            },
+        }
+
+    if verbose:
+        print(f"\n--- PROMPT ({model}) ---")
+        print(prompt[:500] + ("..." if len(prompt) > 500 else ""))
+        print("--- FIN PROMPT ---\n")
+
+    for attempt in range(2):
+        try:
+            t0 = time.time()
+            response = requests.post(endpoint, json=body, timeout=timeout)
+            elapsed = time.time() - t0
+            response.raise_for_status()
+            data = response.json()
+
+            # Extraire le texte de la réponse selon l'API utilisée
+            if is_qwen:
+                raw = data.get("message", {}).get("content", "")
+            else:
+                raw = data.get("response", "")
+
+            if verbose:
+                print(f"--- RÉPONSE ({elapsed:.1f}s) ---")
+                print(raw[:500] + ("..." if len(raw) > 500 else ""))
+                print("--- FIN RÉPONSE ---\n")
+
+            result = parse_json_response(raw)
+            if result is not None:
+                return result
+            if attempt == 0:
+                print(f"  [warn] JSON invalide, retry... (raw: {raw[:100]})")
+        except requests.ConnectionError:
+            print("[ERREUR] Ollama non disponible sur localhost:11434")
+            sys.exit(1)
+        except requests.Timeout:
+            print(f"  [warn] Timeout ({timeout}s) — tentative {attempt + 1}/2")
+            if attempt == 1:
+                return None
+        except requests.RequestException as e:
+            print(f"  [warn] Erreur requête : {e}")
+            return None
+
+    return None
+
+
+# ---------------------------------------------------------------------------
+# 3. Phase 1 — Détection du type de document
+# ---------------------------------------------------------------------------
+
+PROMPT_PHASE1 = """\
+Tu es un expert en codage PMSI et contrôle T2A. Analyse le début de ce document et identifie sa structure.
+
+TEXTE (début du document) :
+---
+{text_preview}
+---
+
+Réponds UNIQUEMENT en JSON avec ces champs :
+{{
+  "type_document": "decision_ucr | notification_cpam | rapport_controle | autre",
+  "organisme": "nom de l'organisme (CPAM, UCR, ARS...)",
+  "date_document": "date au format YYYY-MM-DD si trouvée, sinon vide",
+  "objet": "résumé en une phrase de l'objet du document",
+  "separateur_blocs": "regex Python pour séparer les dossiers individuels (ex: OGC \\\\d+ :)",
+  "colonnes_detectees": ["liste des champs/colonnes détectés dans la structure"]
+}}
+
+IMPORTANT :
+- Le separateur_blocs doit être un regex Python valide
+- Il doit capturer le motif qui sépare chaque dossier/cas individuel
+- Si c'est un document UCR, le séparateur est typiquement "OGC \\\\d+ :"
+- Si tu ne trouves pas de séparateur clair, mets une chaîne vide ""
+"""
+
+
+def detect_document_type(full_text: str, model: str, timeout: int, verbose: bool) -> dict:
+    """Phase 1 : détection du type de document via LLM."""
+    preview = full_text[:3000]
+    prompt = PROMPT_PHASE1.format(text_preview=preview)
+    result = call_ollama(prompt, model=model, timeout=timeout, verbose=verbose)
+    if result is None:
+        print("  [warn] Phase 1 : détection échouée, utilisation des valeurs par défaut")
+        return {
+            "type_document": "autre",
+            "organisme": "",
+            "date_document": "",
+            "objet": "",
+            "separateur_blocs": "",
+            "colonnes_detectees": [],
+        }
+    return result
+
+
+# ---------------------------------------------------------------------------
+# 4. Découpage en blocs
+# ---------------------------------------------------------------------------
+
+def split_into_blocks(full_text: str, separator_pattern: str) -> list[str]:
+    """Découpe le texte en blocs logiques (dossiers individuels)."""
+    blocks = []
+
+    # Tentative avec le séparateur détecté par le LLM
+    if separator_pattern:
+        try:
+            regex = re.compile(separator_pattern, re.MULTILINE | re.IGNORECASE)
+            parts = regex.split(full_text)
+            # Recombiner : le séparateur fait partie du bloc suivant
+            matches = list(regex.finditer(full_text))
+            if len(matches) >= 3:
+                for i, match in enumerate(matches):
+                    start = match.start()
+                    end = matches[i + 1].start() if i + 1 < len(matches) else len(full_text)
+                    block = full_text[start:end].strip()
+                    if block:
+                        blocks.append(block)
+                print(f"  Découpage par séparateur : {len(blocks)} blocs trouvés")
+                return blocks
+            else:
+                print(f"  [warn] Séparateur '{separator_pattern}' → seulement {len(matches)} blocs, fallback")
+        except re.error as e:
+            print(f"  [warn] Regex invalide '{separator_pattern}' : {e}, fallback")
+
+    # Fallback : découpage par taille (~6000 chars, chevauchement 500)
+    chunk_size = 6000
+    overlap = 500
+    text_len = len(full_text)
+    if text_len <= chunk_size:
+        return [full_text]
+
+    pos = 0
+    while pos < text_len:
+        end = min(pos + chunk_size, text_len)
+        # Essayer de couper à une fin de ligne
+        if end < text_len:
+            newline_pos = full_text.rfind("\n", pos + chunk_size - 200, end + 200)
+            if newline_pos > pos:
+                end = newline_pos
+        blocks.append(full_text[pos:end].strip())
+        pos = end - overlap if end < text_len else text_len
+
+    print(f"  Découpage par taille : {len(blocks)} blocs ({chunk_size} chars, chevauchement {overlap})")
+    return blocks
+
+
+# ---------------------------------------------------------------------------
+# 5. Phase 2 — Extraction bloc par bloc
+# ---------------------------------------------------------------------------
+
+SCHEMA_FIELDS = """\
+Champs à extraire (JSON) — remplis chaque champ ou laisse une chaîne vide "" si non trouvé :
+- "champ": numéro de champ (entier, 0 si non trouvé)
+- "ogc": numéro OGC / numéro de dossier (entier, 0 si non trouvé)
+- "type_desaccord": type de désaccord — "DP", "DAS", "DP + DAS", ou ""
+- "code_etablissement": code(s) CIM-10 de l'établissement (ex: "G40.0 + F10.2")
+- "libelle_etablissement": libellé(s) correspondant aux codes établissement
+- "code_controleurs": code(s) CIM-10 des contrôleurs (ou "non repris")
+- "libelle_controleurs": libellé(s) correspondant aux codes contrôleurs
+- "codes_retenus_final": code(s) finalement retenus par l'UCR/la décision
+- "decision": classification — "Favorable établissement", "Défavorable établissement", "Mixte", ou "Indéterminé"
+  * "Favorable établissement" = la décision retient l'avis/le codage de l'établissement
+  * "Défavorable établissement" = la décision confirme l'avis des contrôleurs
+  * "Mixte" = partiellement favorable et partiellement défavorable
+  * "Indéterminé" = impossible à classifier clairement
+- "texte_decision_complet": texte intégral de la décision/conclusion
+- "resume_motif": résumé en 1-2 phrases du motif de la décision
+- "regles_citees": règles de codage citées (ex: "T3, T7")
+- "references_guide": références documentaires (guide méthodologique, fascicules ATIH, avis Agora…)
+- "ghm_mentionne": tous les GHM mentionnés (ex: "05M09 / 05M092")
+- "ghs_mentionne": tous les GHS mentionnés
+- "ghm_final": le GHM final retenu
+- "ghs_final": le GHS final retenu
+- "impact_groupage": impact sur le groupage — "Mieux valorisé", "Pas de changement", ou ""
+"""
+
+PROMPT_PHASE2 = """\
+Tu es un expert en codage PMSI et contrôle T2A.
+
+CONTEXTE DOCUMENT :
+- Type : {type_document}
+- Organisme : {organisme}
+- Objet : {objet}
+
+BLOC DE TEXTE À ANALYSER :
+---
+{block_text}
+---
+
+CONSIGNES :
+1. Extrais les informations de chaque dossier/cas présent dans ce bloc.
+2. Si le bloc contient UN SEUL dossier, retourne un objet JSON.
+3. Si le bloc contient PLUSIEURS dossiers, retourne une LISTE d'objets JSON.
+4. Si le bloc ne contient aucun dossier exploitable (en-tête, pied de page, texte administratif sans cas individuel), retourne : {{"skip": true}}
+
+{schema}
+
+IMPORTANT :
+- Sois précis sur les codes CIM-10 (format X00.0)
+- Pour la décision, analyse attentivement le texte : "retient l'avis de l'établissement" = Favorable, "confirme l'avis des contrôleurs" = Défavorable
+- Ne laisse aucun champ sans clé, utilise "" pour les valeurs inconnues
+- Retourne UNIQUEMENT du JSON valide, sans texte avant ou après
+"""
+
+
+def extract_block(
+    block_text: str,
+    doc_info: dict,
+    model: str,
+    timeout: int,
+    verbose: bool,
+) -> list[dict]:
+    """Extrait les données d'un bloc via LLM. Retourne une liste de dossiers."""
+    prompt = PROMPT_PHASE2.format(
+        type_document=doc_info.get("type_document", "autre"),
+        organisme=doc_info.get("organisme", ""),
+        objet=doc_info.get("objet", ""),
+        block_text=block_text[:8000],  # Limiter la taille
+        schema=SCHEMA_FIELDS,
+    )
+    result = call_ollama(prompt, model=model, max_tokens=4000, timeout=timeout, verbose=verbose)
+    if result is None:
+        return []
+
+    # Skip
+    if isinstance(result, dict) and result.get("skip"):
+        return []
+
+    # Normaliser en liste
+    if isinstance(result, dict):
+        items = [result]
+    elif isinstance(result, list):
+        items = [r for r in result if isinstance(r, dict) and not r.get("skip")]
+    else:
+        return []
+
+    return items
+
+
+# ---------------------------------------------------------------------------
+# 6. Fusion et dédoublonnage
+# ---------------------------------------------------------------------------
+
+# Mapping clés LLM (snake_case) → clés Excel (TitleCase)
+KEY_MAP = {
+    "champ": "Champ",
+    "ogc": "OGC",
+    "type_desaccord": "Type_desaccord",
+    "code_etablissement": "Code_etablissement",
+    "libelle_etablissement": "Libelle_etablissement",
+    "code_controleurs": "Code_controleurs",
+    "libelle_controleurs": "Libelle_controleurs",
+    "codes_retenus_final": "Codes_retenus_final",
+    "decision": "Decision",
+    "texte_decision_complet": "Texte_decision_complet",
+    "resume_motif": "Resume_motif",
+    "regles_citees": "Regles_citees",
+    "references_guide": "References_guide",
+    "ghm_mentionne": "GHM_mentionne",
+    "ghs_mentionne": "GHS_mentionne",
+    "ghm_final": "GHM_final",
+    "ghs_final": "GHS_final",
+    "impact_groupage": "Impact_groupage",
+}
+
+
+def normalize_row(raw: dict) -> dict:
+    """Convertit les clés LLM en clés Excel et normalise les types."""
+    row = {}
+    for llm_key, excel_key in KEY_MAP.items():
+        val = raw.get(llm_key, raw.get(excel_key, ""))
+        # Convertir en int pour Champ et OGC
+        if excel_key in ("Champ", "OGC"):
+            try:
+                val = int(val) if val else 0
+            except (ValueError, TypeError):
+                val = 0
+        elif not isinstance(val, str):
+            val = str(val) if val is not None else ""
+        row[excel_key] = val
+    return row
+
+
+def merge_and_deduplicate(all_items: list[dict]) -> list[dict]:
+    """Fusionne, déduplique par OGC, et trie les résultats."""
+    rows = [normalize_row(item) for item in all_items]
+
+    # Filtrer les lignes sans contenu utile
+    rows = [r for r in rows if r["OGC"] > 0 or r["Code_etablissement"] or r["Decision"]]
+
+    # Dédoublonnage par OGC (garder la version la plus complète)
+    seen: dict[int, dict] = {}
+    deduped: list[dict] = []
+    for r in rows:
+        key = r["OGC"]
+        if key == 0:
+            deduped.append(r)
+            continue
+        if key in seen:
+            old = seen[key]
+            old_score = sum(1 for v in old.values() if v and v != 0)
+            new_score = sum(1 for v in r.values() if v and v != 0)
+            if new_score > old_score:
+                deduped = [x for x in deduped if x["OGC"] != key]
+                deduped.append(r)
+                seen[key] = r
+        else:
+            seen[key] = r
+            deduped.append(r)
+
+    deduped.sort(key=lambda r: (r["Champ"], r["OGC"]))
+    return deduped
+
+
+# ---------------------------------------------------------------------------
+# 7. Export Excel
+# ---------------------------------------------------------------------------
+
+HEADERS = [
+    "Champ", "OGC", "Type_desaccord",
+    "Code_etablissement", "Libelle_etablissement",
+    "Code_controleurs", "Libelle_controleurs",
+    "Codes_retenus_final",
+    "Decision", "Texte_decision_complet", "Resume_motif",
+    "Regles_citees", "References_guide",
+    "GHM_mentionne", "GHS_mentionne", "GHM_final", "GHS_final",
+    "Impact_groupage",
+]
+
+HEADER_LABELS = [
+    "Champ", "N° OGC", "Type désaccord",
+    "Code(s) Établissement", "Libellé Établissement",
+    "Code(s) Contrôleurs", "Libellé Contrôleurs",
+    "Code(s) retenus (final)",
+    "Décision UCR", "Texte décision complet", "Résumé du motif",
+    "Règles codage citées", "Références (guide, fascicules, avis)",
+    "GHM mentionné(s)", "GHS mentionné(s)", "GHM final", "GHS final",
+    "Impact groupage",
+]
+
+
+def write_excel(rows: list[dict], output_path: str):
+    """Écrit les résultats dans un fichier Excel (feuille unique)."""
+    wb = Workbook()
+    ws = wb.active
+    ws.title = "Décisions UCR"
+
+    # Styles
+    header_font = Font(bold=True, color="FFFFFF", size=11)
+    header_fill = PatternFill(start_color="2F5496", end_color="2F5496", fill_type="solid")
+    header_align = Alignment(horizontal="center", vertical="center", wrap_text=True)
+    thin_border = Border(
+        left=Side(style="thin"), right=Side(style="thin"),
+        top=Side(style="thin"), bottom=Side(style="thin"),
+    )
+
+    fav_fill = PatternFill(start_color="C6EFCE", end_color="C6EFCE", fill_type="solid")
+    defav_fill = PatternFill(start_color="FFC7CE", end_color="FFC7CE", fill_type="solid")
+    mixte_fill = PatternFill(start_color="FFEB9C", end_color="FFEB9C", fill_type="solid")
+
+    # En-têtes
+    for col, label in enumerate(HEADER_LABELS, 1):
+        cell = ws.cell(row=1, column=col, value=label)
+        cell.font = header_font
+        cell.fill = header_fill
+        cell.alignment = header_align
+        cell.border = thin_border
+
+    # Données
+    for row_idx, data in enumerate(rows, 2):
+        for col_idx, key in enumerate(HEADERS, 1):
+            val = data.get(key, "")
+            cell = ws.cell(row=row_idx, column=col_idx, value=val)
+            cell.border = thin_border
+            cell.alignment = Alignment(vertical="top", wrap_text=True)
+
+        # Colorer la colonne Décision
+        dec_col = HEADERS.index("Decision") + 1
+        decision_cell = ws.cell(row=row_idx, column=dec_col)
+        dv = str(decision_cell.value or "")
+        if "Favorable" in dv and "Défavorable" not in dv:
+            decision_cell.fill = fav_fill
+        elif "Défavorable" in dv:
+            decision_cell.fill = defav_fill
+        elif "Mixte" in dv:
+            decision_cell.fill = mixte_fill
+
+    # Largeurs de colonnes
+    col_widths = {
+        "Champ": 8, "OGC": 8, "Type_desaccord": 14,
+        "Code_etablissement": 22, "Libelle_etablissement": 40,
+        "Code_controleurs": 22, "Libelle_controleurs": 40,
+        "Codes_retenus_final": 22,
+        "Decision": 24, "Texte_decision_complet": 80,
+        "Resume_motif": 60,
+        "Regles_citees": 16, "References_guide": 50,
+        "GHM_mentionne": 16, "GHS_mentionne": 16,
+        "GHM_final": 12, "GHS_final": 10,
+        "Impact_groupage": 20,
+    }
+    for i, key in enumerate(HEADERS, 1):
+        ws.column_dimensions[ws.cell(row=1, column=i).column_letter].width = col_widths.get(key, 15)
+
+    # Filtre automatique + freeze
+    last_col_letter = ws.cell(row=1, column=len(HEADERS)).column_letter
+    ws.auto_filter.ref = f"A1:{last_col_letter}{len(rows)+1}"
+    ws.freeze_panes = "A2"
+
+    wb.save(output_path)
+    print(f"Excel enregistré : {output_path}")
+
+
+# ---------------------------------------------------------------------------
+# 8. CLI / Main
+# ---------------------------------------------------------------------------
+
+def main():
+    parser = argparse.ArgumentParser(
+        description="Extracteur T2A généraliste via OCR + LLM (Ollama)",
+    )
+    parser.add_argument("pdf", help="Fichier PDF à traiter")
+    parser.add_argument("--model", default="gemma3:27b-it-qat",
+                        help="Modèle Ollama (défaut: gemma3:27b-it-qat)")
+    parser.add_argument("--timeout", type=int, default=120,
+                        help="Timeout par appel LLM en secondes (défaut: 120)")
+    parser.add_argument("--output", default=None,
+                        help="Fichier Excel de sortie (défaut: <nom>_llm.xlsx)")
+    parser.add_argument("--dpi", type=int, default=300,
+                        help="Résolution OCR (défaut: 300)")
+    parser.add_argument("--no-cache", action="store_true",
+                        help="Désactiver le cache texte OCR")
+    parser.add_argument("--verbose", action="store_true",
+                        help="Afficher les prompts/réponses LLM")
+
+    args = parser.parse_args()
+
+    pdf_path = args.pdf
+    if not Path(pdf_path).exists():
+        print(f"[ERREUR] Fichier non trouvé : {pdf_path}")
+        sys.exit(1)
+
+    output_path = args.output or str(Path(pdf_path).with_name(
+        Path(pdf_path).stem + "_llm.xlsx"
+    ))
+
+    print(f"Fichier PDF   : {pdf_path}")
+    print(f"Modèle LLM   : {args.model}")
+    print(f"Sortie Excel  : {output_path}")
+    print()
+
+    # --- Étape 1 : OCR ---
+    txt_cache = Path(pdf_path).with_suffix(".txt")
+    if txt_cache.exists() and not args.no_cache:
+        print("Étape 1/4 : Chargement du texte depuis le cache...")
+        full_text = txt_cache.read_text(encoding="utf-8")
+        full_text = normalize_text(full_text)
+        print(f"  {len(full_text)} caractères chargés depuis {txt_cache}")
+    else:
+        print("Étape 1/4 : OCR du document...")
+        full_text = ocr_pdf(pdf_path, dpi=args.dpi)
+        if not args.no_cache:
+            txt_cache.write_text(full_text, encoding="utf-8")
+            print(f"  Cache texte sauvegardé : {txt_cache}")
+    print(f"  Longueur du texte : {len(full_text)} caractères")
+    print()
+
+    # --- Étape 2 : Détection du type de document ---
+    print("Étape 2/4 : Détection du type de document...")
+    t0 = time.time()
+    doc_info = detect_document_type(full_text, model=args.model, timeout=args.timeout, verbose=args.verbose)
+    print(f"  Type      : {doc_info.get('type_document', '?')}")
+    print(f"  Organisme : {doc_info.get('organisme', '?')}")
+    print(f"  Objet     : {doc_info.get('objet', '?')}")
+    print(f"  Séparateur: {doc_info.get('separateur_blocs', '(aucun)')}")
+    print(f"  Colonnes  : {doc_info.get('colonnes_detectees', [])}")
+    print(f"  ({time.time() - t0:.1f}s)")
+    print()
+
+    # --- Étape 3 : Découpage et extraction ---
+    print("Étape 3/4 : Découpage en blocs et extraction LLM...")
+    separator = doc_info.get("separateur_blocs", "")
+    blocks = split_into_blocks(full_text, separator)
+    print(f"  {len(blocks)} blocs à traiter")
+
+    all_items = []
+    t0 = time.time()
+    for i, block in enumerate(blocks):
+        print(f"  Bloc {i+1}/{len(blocks)}...", end="\r")
+        items = extract_block(block, doc_info, model=args.model, timeout=args.timeout, verbose=args.verbose)
+        all_items.extend(items)
+        # Progression
+        elapsed = time.time() - t0
+        avg = elapsed / (i + 1)
+        remaining = avg * (len(blocks) - i - 1)
+        print(f"  Bloc {i+1}/{len(blocks)} → {len(items)} dossier(s)  "
+              f"[{elapsed:.0f}s écoulé, ~{remaining:.0f}s restant]   ")
+
+    total_elapsed = time.time() - t0
+    print(f"  Extraction terminée : {len(all_items)} dossiers bruts en {total_elapsed:.0f}s")
+    print()
+
+    # --- Étape 4 : Fusion et export ---
+    print("Étape 4/4 : Fusion, dédoublonnage et export Excel...")
+    rows = merge_and_deduplicate(all_items)
+    print(f"  {len(rows)} dossiers après dédoublonnage")
+
+    # Statistiques
+    fav = sum(1 for r in rows if "Favorable" in r.get("Decision", "") and "Défavorable" not in r.get("Decision", ""))
+    defav = sum(1 for r in rows if "Défavorable" in r.get("Decision", ""))
+    mixte = sum(1 for r in rows if "Mixte" in r.get("Decision", ""))
+    indet = sum(1 for r in rows if r.get("Decision", "") in ("Indéterminé", ""))
+    print(f"  Favorable établissement  : {fav}")
+    print(f"  Défavorable établissement : {defav}")
+    print(f"  Mixte                     : {mixte}")
+    print(f"  Indéterminé               : {indet}")
+
+    write_excel(rows, output_path)
+    print()
+    print("Terminé.")
+
+
+if __name__ == "__main__":
+    main()

cpam/parse_decision_ucr.py

@@ -0,0 +1,690 @@
+#!/usr/bin/env python3
+"""
+parse_decision_ucr.py — Extraction des décisions UCR depuis un PDF scanné (contrôle T2A)
+
+Entrée  : PDF scanné de décision UCR (CPAM / Assurance Maladie)
+Sortie  : Fichier Excel (.xlsx) avec une feuille unique
+
+Colonnes extraites (enrichies pour analyse IA) :
+  Champ, OGC, Type_desaccord,
+  Code_etablissement, Libelle_etablissement,
+  Code_controleurs, Libelle_controleurs,
+  Codes_retenus_final,
+  Decision, Texte_decision_complet, Resume_motif,
+  Regles_citees, References_guide,
+  GHM_mentionne, GHS_mentionne, GHM_final, GHS_final,
+  Impact_groupage
+"""
+from __future__ import annotations
+
+import re
+import sys
+from pathlib import Path
+
+import pymupdf
+import pytesseract
+from PIL import Image
+import io
+from openpyxl import Workbook
+from openpyxl.styles import Font, PatternFill, Alignment, Border, Side
+import unicodedata
+
+
+# ---------------------------------------------------------------------------
+# 0. Normalisation texte OCR
+# ---------------------------------------------------------------------------
+
+def normalize_text(text: str) -> str:
+    """Normalise les apostrophes, guillemets et espaces issus de l'OCR."""
+    text = text.replace("\u2018", "'").replace("\u2019", "'")
+    text = text.replace("\u201C", '"').replace("\u201D", '"')
+    text = text.replace("\u00AB", '"').replace("\u00BB", '"')
+    text = text.replace("''", "'")
+    text = text.replace("\u00A0", " ").replace("\u202F", " ")
+    # Erreurs OCR courantes
+    text = re.sub(r"\bF'UCR\b", "l'UCR", text)
+    text = re.sub(r"\bl''UCR\b", "l'UCR", text)
+    return text
+
+
+# ---------------------------------------------------------------------------
+# 1. OCR
+# ---------------------------------------------------------------------------
+
+def ocr_pdf(pdf_path: str, dpi: int = 300) -> str:
+    """Extrait le texte de toutes les pages du PDF via Tesseract OCR."""
+    doc = pymupdf.open(pdf_path)
+    full_text = []
+    total = len(doc)
+    for i, page in enumerate(doc):
+        print(f"  OCR page {i+1}/{total}...", end="\r")
+        mat = pymupdf.Matrix(dpi / 72, dpi / 72)
+        pix = page.get_pixmap(matrix=mat)
+        img = Image.open(io.BytesIO(pix.tobytes("png")))
+        text = pytesseract.image_to_string(img, lang="fra")
+        full_text.append(text)
+    print(f"  OCR terminé : {total} pages.          ")
+    return normalize_text("\n\n".join(full_text))
+
+
+# ---------------------------------------------------------------------------
+# 2. Parsing — Regex
+# ---------------------------------------------------------------------------
+
+RE_CHAMP = re.compile(
+    r"Champ\s*(?:n°\s*)?(\d+)\s*[:\-—]?\s*(?:Séjours|:)",
+    re.IGNORECASE,
+)
+
+RE_OGC_HEADER = re.compile(
+    r"(?:^|\n)\s*OGC\s+(\d+)\s*:",
+    re.MULTILINE,
+)
+
+RE_TYPE_DESACCORD = re.compile(
+    r"(?:désaccord|discussion)\s+porte\s+(?:sur\s+)?(?:le\s+|les\s+)?(DP\s+et\s+(?:le\s+)?DAS|DP\s+et\s+DAS|DP|DAS)",
+    re.IGNORECASE,
+)
+
+RE_CIM10 = re.compile(r"\b([A-Z]\d{2}(?:\.\d{1,2})?)\b")
+
+RE_CODAGE_ETS = re.compile(
+    r"Codage\s+[ée]tablissement\s*:\s*(.*?)(?=Codage\s+contr[ôo]leurs)",
+    re.IGNORECASE | re.DOTALL,
+)
+
+RE_CODAGE_CTRL = re.compile(
+    r"Codage\s+contr[ôo]leurs\s*:\s*(.*?)(?=D[EÉ]C[I1]?SION\s+UCR|PROPOSITION\s+UCR)",
+    re.IGNORECASE | re.DOTALL,
+)
+
+RE_DECISION = re.compile(
+    r"(?:D[EÉ]C[I1]?SION|PROPOSITION)\s+UCR\s*:?\s*(.*)",
+    re.IGNORECASE | re.DOTALL,
+)
+
+# --- Classification ---
+
+RE_FAVORABLE = re.compile(
+    r"(?:"
+    r"retient\s+(?:la\s+demande|le\s+codage|l'avis)\s+(?:de\s+)?l'[ée]tablissement"
+    r"|retient\s+en\s+D[PA]S\s+le\s+code"
+    r"|retient\s+le\s+codage\s+du\s+DP\s+de\s+l'[ée]tablissement"
+    r"|l'UCR\s+retient\s+l'avis\s+de\s+l'[ée]tablissement"
+    r"|confirme\s+l'avis\s+(?:de\s+)?l'[ée]tablissement"
+    r")",
+    re.IGNORECASE,
+)
+
+RE_DEFAVORABLE = re.compile(
+    r"confirme\s+l'avis\s+des\s+(?:m[ée]decins\s+)?contr[oô]leurs",
+    re.IGNORECASE,
+)
+
+RE_UCR_RETIENT = re.compile(r"l'UCR\s+retient\b", re.IGNORECASE)
+RE_UCR_PROPOSE = re.compile(r"l'UCR\s+propose\b", re.IGNORECASE)
+RE_NE_RETIENT_PAS = re.compile(r"ne\s+retient\s+pas", re.IGNORECASE)
+
+# --- GHM / GHS ---
+
+RE_GHM = re.compile(r"GHM\s+([A-Z0-9]{5,7})", re.IGNORECASE)
+RE_GHS = re.compile(r"GHS\s+(\d{3,5})", re.IGNORECASE)
+
+RE_MIEUX_VALORISE = re.compile(r"mieux\s+valoris[ée]", re.IGNORECASE)
+RE_PAS_MODIFIE = re.compile(
+    r"(?:ne\s+modifie\s+pas|ne\s+change(?:nt)?\s+pas|pas\s+de\s+changement|reste\s+group[ée])",
+    re.IGNORECASE,
+)
+
+# --- Règles et références ---
+
+# Pages du guide méthodologique
+RE_GUIDE_PAGE = re.compile(
+    r"(?:guide\s+m[ée]thodologique|guide)\s*(?:p\.?|page)\s*(\d{1,3})",
+    re.IGNORECASE,
+)
+RE_PAGE_GUIDE = re.compile(
+    r"(?:p\.?|page)\s*(\d{1,3})\s+du\s+guide",
+    re.IGNORECASE,
+)
+
+# Règles T (T3, T7, etc.)
+RE_REGLE_T = re.compile(
+    r"r[èe]gle\s+(T\d+)",
+    re.IGNORECASE,
+)
+
+# Fascicules ATIH
+RE_FASCICULE = re.compile(
+    r"fascicule\s+(?:ATIH\s+)?(?:de\s+codage\s+)?(?:PMSI\s+)?(?:n°\s*)?(\d{1,2})?\s*(?:[-–]\s*)?([A-ZÀ-Üa-zà-ü\s]+?)(?:\s+(?:de\s+)?(\d{4}))?(?:\s*(?:,\s*)?(?:p\.?\s*|page\s*)(\d+))?",
+    re.IGNORECASE,
+)
+
+# Avis Agora
+RE_AVIS_AGORA = re.compile(
+    r"avis\s+(?:agora|AGORA)\s*(?:n°\s*)?(\d+)",
+    re.IGNORECASE,
+)
+
+# Consignes de codage avec page
+RE_CONSIGNES_CODAGE = re.compile(
+    r"consignes?\s+de\s+codage\s*(?:p\.?\s*|page\s*)(\d+)",
+    re.IGNORECASE,
+)
+
+# Codage retenu / DP retenu / DAS retenu
+RE_CODAGE_RETENU = re.compile(
+    r"(?:codage\s+retenu|DP\s*(?:retenu|=)|DAS\s*(?:retenu|=)|code\s+retenu|est\s+cod[ée]\s+en|se\s+code)\s*(?:est\s+)?(?::?\s*)([A-Z]\d{2}(?:\.\d{1,2})?)",
+    re.IGNORECASE,
+)
+
+# "est ajouté en DAS" / "ajout du code X"
+RE_CODE_AJOUTE = re.compile(
+    r"(?:est\s+ajout[ée]\s+en\s+D[PA]S|ajout(?:er)?\s+(?:du\s+|en\s+D[PA]S\s+(?:le\s+)?)?(?:code\s+)?)\s*(?::?\s*)([A-Z]\d{2}(?:\.\d{1,2})?)",
+    re.IGNORECASE,
+)
+
+
+# ---------------------------------------------------------------------------
+# 2b. Fonctions d'extraction
+# ---------------------------------------------------------------------------
+
+def extract_codes_and_label(text: str) -> tuple[str, str]:
+    """Extrait les codes CIM-10 et le libellé depuis un bloc de codage."""
+    codes = RE_CIM10.findall(text)
+    labels = re.findall(r'[«"](.*?)[»"]', text)
+    code_str = " + ".join(codes) if codes else ""
+    label_str = " | ".join(labels) if labels else text.strip()[:120]
+    label_str = re.sub(r"\s+", " ", label_str).strip()
+    return code_str, label_str
+
+
+def extract_codes_retenus(decision_text: str) -> str:
+    """Extrait les codes finalement retenus par l'UCR."""
+    codes = set()
+    for m in RE_CODAGE_RETENU.finditer(decision_text):
+        codes.add(m.group(1))
+    for m in RE_CODE_AJOUTE.finditer(decision_text):
+        codes.add(m.group(1))
+    return " + ".join(sorted(codes)) if codes else ""
+
+
+def extract_regles(text: str) -> str:
+    """Extrait les règles de codage citées (T3, T7, etc.)."""
+    regles = set()
+    for m in RE_REGLE_T.finditer(text):
+        regles.add(m.group(1).upper())
+    return ", ".join(sorted(regles)) if regles else ""
+
+
+def extract_references(text: str) -> str:
+    """Extrait toutes les références (guide, fascicules, avis Agora, consignes)."""
+    refs = []
+
+    # Pages du guide méthodologique
+    pages_guide = set()
+    for m in RE_GUIDE_PAGE.finditer(text):
+        pages_guide.add(m.group(1))
+    for m in RE_PAGE_GUIDE.finditer(text):
+        pages_guide.add(m.group(1))
+    if pages_guide:
+        refs.append("Guide méthodologique p." + ", p.".join(sorted(pages_guide, key=int)))
+
+    # Fascicules ATIH
+    for m in RE_FASCICULE.finditer(text):
+        num = m.group(1) or ""
+        sujet = (m.group(2) or "").strip()
+        annee = m.group(3) or ""
+        page = m.group(4) or ""
+        ref = "Fascicule"
+        if num:
+            ref += f" {num}"
+        if sujet:
+            ref += f" {sujet}"
+        if annee:
+            ref += f" ({annee})"
+        if page:
+            ref += f" p.{page}"
+        refs.append(ref.strip())
+
+    # Avis Agora
+    for m in RE_AVIS_AGORA.finditer(text):
+        refs.append(f"Avis Agora n°{m.group(1)}")
+
+    # Consignes de codage
+    for m in RE_CONSIGNES_CODAGE.finditer(text):
+        refs.append(f"Consignes de codage p.{m.group(1)}")
+
+    # Dédupliquer
+    seen = set()
+    unique = []
+    for r in refs:
+        r_lower = r.lower()
+        if r_lower not in seen:
+            seen.add(r_lower)
+            unique.append(r)
+
+    return " ; ".join(unique) if unique else ""
+
+
+def extract_ghm_ghs_all(text: str) -> tuple[list[str], list[str]]:
+    """Extrait tous les GHM et GHS mentionnés."""
+    ghms = []
+    for m in RE_GHM.finditer(text):
+        v = m.group(1).upper()
+        if v not in ghms:
+            ghms.append(v)
+    ghss = []
+    for m in RE_GHS.finditer(text):
+        v = m.group(1)
+        if v not in ghss:
+            ghss.append(v)
+    return ghms, ghss
+
+
+def classify_decision(decision_text: str) -> str:
+    """Classifie la décision : Favorable / Défavorable / Mixte / Indéterminé."""
+    text = normalize_text(decision_text)
+
+    fav = bool(RE_FAVORABLE.search(text))
+    defav = bool(RE_DEFAVORABLE.search(text))
+
+    ucr_retient = bool(RE_UCR_RETIENT.search(text))
+    ucr_propose = bool(RE_UCR_PROPOSE.search(text))
+    ne_retient_pas = bool(RE_NE_RETIENT_PAS.search(text))
+
+    if ucr_retient and not ne_retient_pas:
+        fav = True
+    if ucr_propose and not defav:
+        fav = True
+
+    if (ucr_retient or fav) and defav:
+        return "Mixte"
+    if fav and defav:
+        return "Mixte"
+    elif fav:
+        return "Favorable établissement"
+    elif defav:
+        return "Défavorable établissement"
+    else:
+        return "Indéterminé"
+
+
+def clean_decision_text(text: str) -> str:
+    """Nettoie le texte de décision (supprime artifacts OCR en fin de bloc)."""
+    # Supprimer les lignes de pied de page UCR
+    text = re.sub(r"\n\s*(?:UCR\s+NA|CONFIDENTIEL|Page\s+\d+).*$", "", text, flags=re.MULTILINE | re.IGNORECASE)
+    # Supprimer les artefacts OCR de fin (séquences de caractères isolés)
+    text = re.sub(r"\n\s*[A-Z]{1,4}\s*(?:—|—|-)\s*[a-zA-Z]{0,3}\s*$", "", text, flags=re.MULTILINE)
+    text = re.sub(r"\n\s*(?:EE|ESS|2 ae|A D ES|EE nd)\s*$", "", text, flags=re.MULTILINE | re.IGNORECASE)
+    # Normaliser les espaces
+    text = re.sub(r"[ \t]+", " ", text)
+    text = re.sub(r"\n{3,}", "\n\n", text)
+    return text.strip()
+
+
+# ---------------------------------------------------------------------------
+# 2c. Parsing des blocs
+# ---------------------------------------------------------------------------
+
+def parse_ogc_block(block_text: str, champ: int, ogc_num: int) -> dict:
+    """Parse un bloc OGC et retourne un dictionnaire structuré enrichi."""
+    result = {
+        "Champ": champ,
+        "OGC": ogc_num,
+        "Type_desaccord": "",
+        "Code_etablissement": "",
+        "Libelle_etablissement": "",
+        "Code_controleurs": "",
+        "Libelle_controleurs": "",
+        "Codes_retenus_final": "",
+        "Decision": "",
+        "Texte_decision_complet": "",
+        "Resume_motif": "",
+        "Regles_citees": "",
+        "References_guide": "",
+        "GHM_mentionne": "",
+        "GHS_mentionne": "",
+        "GHM_final": "",
+        "GHS_final": "",
+        "Impact_groupage": "",
+    }
+
+    # Type de désaccord
+    m = RE_TYPE_DESACCORD.search(block_text)
+    if m:
+        raw = m.group(1).upper().strip()
+        raw = re.sub(r"\s+", " ", raw)
+        if "DP" in raw and "DAS" in raw:
+            result["Type_desaccord"] = "DP + DAS"
+        elif "DAS" in raw:
+            result["Type_desaccord"] = "DAS"
+        elif "DP" in raw:
+            result["Type_desaccord"] = "DP"
+
+    # Codage établissement
+    m = RE_CODAGE_ETS.search(block_text)
+    if m:
+        raw_ets = m.group(1).strip()
+        result["Code_etablissement"], result["Libelle_etablissement"] = extract_codes_and_label(raw_ets)
+
+    # Codage contrôleurs
+    m = RE_CODAGE_CTRL.search(block_text)
+    if m:
+        raw_ctrl = m.group(1).strip()
+        if re.search(r"non\s+repris", raw_ctrl, re.IGNORECASE):
+            result["Code_controleurs"] = "non repris"
+            result["Libelle_controleurs"] = ""
+        else:
+            result["Code_controleurs"], result["Libelle_controleurs"] = extract_codes_and_label(raw_ctrl)
+
+    # Décision UCR — TEXTE COMPLET
+    m = RE_DECISION.search(block_text)
+    if m:
+        decision_text = m.group(1).strip()
+        decision_clean = clean_decision_text(decision_text)
+
+        result["Decision"] = classify_decision(decision_clean)
+        result["Texte_decision_complet"] = decision_clean
+
+        # Résumé court (première phrase significative)
+        resume = re.sub(r"\s+", " ", decision_clean)[:300].strip()
+        # Couper à la dernière phrase complète
+        last_dot = resume.rfind(".")
+        if last_dot > 100:
+            resume = resume[:last_dot + 1]
+        result["Resume_motif"] = resume
+
+        # Codes finalement retenus
+        result["Codes_retenus_final"] = extract_codes_retenus(decision_clean)
+
+        # Règles citées (T3, T7, etc.)
+        result["Regles_citees"] = extract_regles(block_text)
+
+        # Références (guide, fascicules, avis Agora)
+        result["References_guide"] = extract_references(block_text)
+
+    # GHM / GHS — tous ceux mentionnés et le dernier (= final)
+    ghms, ghss = extract_ghm_ghs_all(block_text)
+    if ghms:
+        result["GHM_mentionne"] = " / ".join(ghms)
+        result["GHM_final"] = ghms[-1]  # Le dernier mentionné est souvent le final
+    if ghss:
+        result["GHS_mentionne"] = " / ".join(ghss)
+        result["GHS_final"] = ghss[-1]
+
+    # Impact groupage
+    if RE_MIEUX_VALORISE.search(block_text):
+        result["Impact_groupage"] = "Mieux valorisé"
+    elif RE_PAS_MODIFIE.search(block_text):
+        result["Impact_groupage"] = "Pas de changement"
+
+    return result
+
+
+def parse_grouped_ogcs(text_block: str, champ: int, ogc_nums: list[int]) -> list[dict]:
+    """Parse un bloc groupé (ex: OGC 14,19,46,50 traités ensemble)."""
+    template = parse_ogc_block(text_block, champ, ogc_nums[0])
+    results = []
+    for num in ogc_nums:
+        row = dict(template)
+        row["OGC"] = num
+        results.append(row)
+    return results
+
+
+def parse_document(full_text: str) -> list[dict]:
+    """Parse le texte OCR complet et retourne la liste des dossiers."""
+    rows = []
+
+    champ_positions = [(m.start(), int(m.group(1))) for m in RE_CHAMP.finditer(full_text)]
+    ogc_positions = [(m.start(), int(m.group(1))) for m in RE_OGC_HEADER.finditer(full_text)]
+
+    def get_champ_for_position(pos: int) -> int:
+        ch = 0
+        for cp, cn in champ_positions:
+            if cp <= pos:
+                ch = cn
+            else:
+                break
+        return ch
+
+    # Blocs groupés
+    RE_GROUPED = re.compile(
+        r"(?:Concernant|Pour)\s+les\s+OGC\s+([\d,\s]+)",
+        re.IGNORECASE,
+    )
+
+    grouped_ogcs = set()
+    for m in RE_GROUPED.finditer(full_text):
+        nums = [int(n.strip()) for n in m.group(1).split(",") if n.strip().isdigit()]
+        if len(nums) > 1:
+            start = m.start()
+            end = len(full_text)
+            for op, on in ogc_positions:
+                if op > start + 50 and on not in nums:
+                    end = op
+                    break
+            block = full_text[start:end]
+            champ = get_champ_for_position(start)
+            group_rows = parse_grouped_ogcs(block, champ, nums)
+            rows.extend(group_rows)
+            grouped_ogcs.update(nums)
+
+    # OGC individuels
+    for idx, (pos, ogc_num) in enumerate(ogc_positions):
+        champ = get_champ_for_position(pos)
+
+        end = len(full_text)
+        for next_pos, _ in ogc_positions[idx + 1:]:
+            if next_pos > pos + 20:
+                end = next_pos
+                break
+        for cp, _ in champ_positions:
+            if pos < cp < end:
+                end = cp
+                break
+
+        block = full_text[pos:end]
+        row = parse_ogc_block(block, champ, ogc_num)
+
+        if ogc_num in grouped_ogcs:
+            if row["Code_etablissement"] and row["Decision"]:
+                rows = [r for r in rows if r["OGC"] != ogc_num]
+                rows.append(row)
+        else:
+            if row["Code_etablissement"] or row["Decision"]:
+                rows.append(row)
+
+    rows.sort(key=lambda r: (r["Champ"], r["OGC"]))
+
+    # Dédupliquer
+    seen = {}
+    deduped = []
+    for r in rows:
+        key = r["OGC"]
+        if key in seen:
+            old = seen[key]
+            old_score = sum(1 for v in old.values() if v)
+            new_score = sum(1 for v in r.values() if v)
+            if new_score > old_score:
+                deduped = [x for x in deduped if x["OGC"] != key]
+                deduped.append(r)
+                seen[key] = r
+        else:
+            seen[key] = r
+            deduped.append(r)
+
+    deduped.sort(key=lambda r: (r["Champ"], r["OGC"]))
+    return deduped
+
+
+# ---------------------------------------------------------------------------
+# 3. Export Excel
+# ---------------------------------------------------------------------------
+
+HEADERS = [
+    "Champ",
+    "OGC",
+    "Type_desaccord",
+    "Code_etablissement",
+    "Libelle_etablissement",
+    "Code_controleurs",
+    "Libelle_controleurs",
+    "Codes_retenus_final",
+    "Decision",
+    "Texte_decision_complet",
+    "Resume_motif",
+    "Regles_citees",
+    "References_guide",
+    "GHM_mentionne",
+    "GHS_mentionne",
+    "GHM_final",
+    "GHS_final",
+    "Impact_groupage",
+]
+
+HEADER_LABELS = [
+    "Champ",
+    "N° OGC",
+    "Type désaccord",
+    "Code(s) Établissement",
+    "Libellé Établissement",
+    "Code(s) Contrôleurs",
+    "Libellé Contrôleurs",
+    "Code(s) retenus (final)",
+    "Décision UCR",
+    "Texte décision complet",
+    "Résumé du motif",
+    "Règles codage citées",
+    "Références (guide, fascicules, avis)",
+    "GHM mentionné(s)",
+    "GHS mentionné(s)",
+    "GHM final",
+    "GHS final",
+    "Impact groupage",
+]
+
+
+def write_excel(rows: list[dict], output_path: str):
+    """Écrit les résultats dans un fichier Excel (feuille unique)."""
+    wb = Workbook()
+    ws = wb.active
+    ws.title = "Décisions UCR"
+
+    # Styles
+    header_font = Font(bold=True, color="FFFFFF", size=11)
+    header_fill = PatternFill(start_color="2F5496", end_color="2F5496", fill_type="solid")
+    header_align = Alignment(horizontal="center", vertical="center", wrap_text=True)
+    thin_border = Border(
+        left=Side(style="thin"),
+        right=Side(style="thin"),
+        top=Side(style="thin"),
+        bottom=Side(style="thin"),
+    )
+
+    fav_fill = PatternFill(start_color="C6EFCE", end_color="C6EFCE", fill_type="solid")
+    defav_fill = PatternFill(start_color="FFC7CE", end_color="FFC7CE", fill_type="solid")
+    mixte_fill = PatternFill(start_color="FFEB9C", end_color="FFEB9C", fill_type="solid")
+
+    # En-têtes
+    for col, label in enumerate(HEADER_LABELS, 1):
+        cell = ws.cell(row=1, column=col, value=label)
+        cell.font = header_font
+        cell.fill = header_fill
+        cell.alignment = header_align
+        cell.border = thin_border
+
+    # Données
+    for row_idx, data in enumerate(rows, 2):
+        for col_idx, key in enumerate(HEADERS, 1):
+            val = data.get(key, "")
+            cell = ws.cell(row=row_idx, column=col_idx, value=val)
+            cell.border = thin_border
+            cell.alignment = Alignment(vertical="top", wrap_text=True)
+
+        # Colorer la colonne Décision
+        dec_col = HEADERS.index("Decision") + 1
+        decision_cell = ws.cell(row=row_idx, column=dec_col)
+        dv = str(decision_cell.value or "")
+        if "Favorable" in dv and "Défavorable" not in dv:
+            decision_cell.fill = fav_fill
+        elif "Défavorable" in dv:
+            decision_cell.fill = defav_fill
+        elif "Mixte" in dv:
+            decision_cell.fill = mixte_fill
+
+    # Largeurs de colonnes
+    col_widths = {
+        "Champ": 8, "OGC": 8, "Type_desaccord": 14,
+        "Code_etablissement": 22, "Libelle_etablissement": 40,
+        "Code_controleurs": 22, "Libelle_controleurs": 40,
+        "Codes_retenus_final": 22,
+        "Decision": 24, "Texte_decision_complet": 80,
+        "Resume_motif": 60,
+        "Regles_citees": 16, "References_guide": 50,
+        "GHM_mentionne": 16, "GHS_mentionne": 16,
+        "GHM_final": 12, "GHS_final": 10,
+        "Impact_groupage": 20,
+    }
+    for i, key in enumerate(HEADERS, 1):
+        ws.column_dimensions[ws.cell(row=1, column=i).column_letter].width = col_widths.get(key, 15)
+
+    # Filtre automatique
+    last_col_letter = ws.cell(row=1, column=len(HEADERS)).column_letter
+    ws.auto_filter.ref = f"A1:{last_col_letter}{len(rows)+1}"
+
+    # Figer la première ligne
+    ws.freeze_panes = "A2"
+
+    wb.save(output_path)
+    print(f"Excel enregistré : {output_path}")
+
+
+# ---------------------------------------------------------------------------
+# Main
+# ---------------------------------------------------------------------------
+
+def main():
+    if len(sys.argv) < 2:
+        pdf_path = str(Path(__file__).parent / "SPHO-FINANC26020915121.pdf")
+    else:
+        pdf_path = sys.argv[1]
+
+    output_path = str(Path(pdf_path).with_suffix(".xlsx"))
+
+    print(f"Fichier PDF : {pdf_path}")
+    print("Étape 1/3 : OCR du document...")
+    full_text = ocr_pdf(pdf_path)
+
+    txt_path = str(Path(pdf_path).with_suffix(".txt"))
+    Path(txt_path).write_text(full_text, encoding="utf-8")
+    print(f"  Texte brut sauvegardé : {txt_path}")
+
+    print("Étape 2/3 : Extraction des décisions...")
+    rows = parse_document(full_text)
+    print(f"  {len(rows)} dossiers OGC extraits.")
+
+    fav = sum(1 for r in rows if "Favorable" in r.get("Decision", "") and "Défavorable" not in r.get("Decision", ""))
+    defav = sum(1 for r in rows if "Défavorable" in r.get("Decision", ""))
+    mixte = sum(1 for r in rows if "Mixte" in r.get("Decision", ""))
+    indet = sum(1 for r in rows if r.get("Decision", "") in ("Indéterminé", ""))
+    refs_count = sum(1 for r in rows if r.get("References_guide"))
+    codes_ret = sum(1 for r in rows if r.get("Codes_retenus_final"))
+    regles = sum(1 for r in rows if r.get("Regles_citees"))
+
+    print(f"  Favorable établissement : {fav}")
+    print(f"  Défavorable établissement : {defav}")
+    print(f"  Mixte : {mixte}")
+    print(f"  Indéterminé : {indet}")
+    print(f"  Avec références citées : {refs_count}")
+    print(f"  Avec codes retenus : {codes_ret}")
+    print(f"  Avec règles T : {regles}")
+
+    print("Étape 3/3 : Génération du fichier Excel...")
+    write_excel(rows, output_path)
+    print("Terminé.")
+
+
+if __name__ == "__main__":
+    main()

data/gold_standard/103_23056749__103_23056749_fusionne_cim10.json

@@ -0,0 +1,198 @@
+{
+  "dossier_id": "103_23056749/103_23056749_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Hallucinations",
+    "code_pipeline": "F06.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Et cardiaque",
+      "code_pipeline": "R00",
+      "confidence": "medium",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Goutte",
+      "code_pipeline": "M10.9",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Syndrome anxio-dépressif",
+      "code_pipeline": "F33.3",
+      "confidence": "medium",
+      "source": "llm_das",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Apragmatisme, aboulie, clinophilie, troubles cognitifs",
+      "code_pipeline": "R48.8",
+      "confidence": "high",
+      "source": "llm_das",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Thrombose veineuse profonde",
+      "code_pipeline": "I80.2",
+      "confidence": "high",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Insuffisance rénale aiguë",
+      "code_pipeline": "N17.8",
+      "confidence": "high",
+      "source": "llm_das",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Maladie de parkinson",
+      "code_pipeline": "G20",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Examen complémentaire",
+      "code_pipeline": "Z04.8",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Anorexie",
+      "code_pipeline": "R63.0",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Troubles du sommeil",
+      "code_pipeline": "F51.0",
+      "confidence": "medium",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Troubles de l'alimentation",
+      "code_pipeline": "F50.9",
+      "confidence": "medium",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Hépatite b",
+      "code_pipeline": "B16.9",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.0",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Hépatite c",
+      "code_pipeline": "B18.2",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Anémie",
+      "code_pipeline": "D50",
+      "confidence": "medium",
+      "source": "llm_das",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 16,
+      "texte_original": "Perte d'autonomie",
+      "code_pipeline": "Z73.6",
+      "confidence": "medium",
+      "source": "llm_das",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 17,
+      "texte_original": "Douleur abdominale",
+      "code_pipeline": "R10",
+      "confidence": "high",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/113_23065949__trackare-18022356-23065949_18022356_23065949_cim10.json

@@ -0,0 +1,108 @@
+{
+  "dossier_id": "113_23065949/trackare-18022356-23065949_18022356_23065949_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Diverticulose du côlon et de l'intestin grêle, (sans perforation ni abcès)",
+    "code_pipeline": "K57.5",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Diverticulose du côlon",
+      "code_pipeline": "K57.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Pathologique",
+      "code_pipeline": "F07.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Colite gauche",
+      "code_pipeline": "K51.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Nausées et vomissements",
+      "code_pipeline": "R11",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Infection post-opératoire/Abcès",
+      "code_pipeline": "T81.8",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Insuffisance rénale aiguë",
+      "code_pipeline": "N17.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/115_23066188__115_23066188_fusionne_cim10.json

@@ -0,0 +1,78 @@
+{
+  "dossier_id": "115_23066188/115_23066188_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Méningite à entérovirus",
+    "code_pipeline": "A87.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Migraine",
+      "code_pipeline": "G43.9",
+      "confidence": "medium",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Céphalées",
+      "code_pipeline": "R51",
+      "confidence": "",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Éruption cutanée médicamenteuse",
+      "code_pipeline": "L27.9",
+      "confidence": "medium",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Céphalées",
+      "code_pipeline": "G44.2",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Constipation",
+      "code_pipeline": "K59.0",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Méningite, sans précision",
+      "code_pipeline": "G03.9",
+      "confidence": "high",
+      "source": "trackare",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/116_23065570__116_23065570_fusionne_cim10.json

@@ -0,0 +1,14 @@
+{
+  "commentaire_general": "",
+  "das": [],
+  "das_ajoutes": [],
+  "dossier_id": "116_23065570/116_23065570_fusionne_cim10",
+  "dp": {
+    "code_corrige": null,
+    "commentaire": "",
+    "statut": "correct"
+  },
+  "statut": "non_commence",
+  "validateur": "",
+  "date_validation": "2026-02-17T21:38:39"
+}

data/gold_standard/118_23042633__118_23042633_fusionne_cim10.json

@@ -0,0 +1,128 @@
+{
+  "dossier_id": "118_23042633/118_23042633_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "X 2 en 2013",
+    "code_pipeline": "X20.13",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Strabisme ",
+      "code_pipeline": "H50.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Examen général",
+      "code_pipeline": "Z00.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Myocardiopathie",
+      "code_pipeline": "I42.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Cervicale",
+      "code_pipeline": "M54.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Cardiopathie ischémique",
+      "code_pipeline": "I25.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Insuffisance rénale chronique modérée",
+      "code_pipeline": "N18.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Lésion du nerf optique droit avec cécité quasi complète",
+      "code_pipeline": "H47.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Hématome",
+      "code_pipeline": "N85.7",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Dissection de l'",
+      "code_pipeline": "I71.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Ventilationventilationventilationventilationventilation",
+      "code_pipeline": "Z99.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/119_23062819__119_23062819_fusionne_cim10.json

@@ -0,0 +1,128 @@
+{
+  "dossier_id": "119_23062819/119_23062819_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Hepatomegalie,",
+    "code_pipeline": "R48.10",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Ascite",
+      "code_pipeline": "R18",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Cryoglobulinémie",
+      "code_pipeline": "D89.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Neuropathie périphérique",
+      "code_pipeline": "G60.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Cyanose",
+      "code_pipeline": "R23.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Examen général",
+      "code_pipeline": "Z00.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Cervicale",
+      "code_pipeline": "M54.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Polype vésical bourgeonnant",
+      "code_pipeline": "C67.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Lésion endovésicale volumineuse",
+      "code_pipeline": "R19.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Pneumonie",
+      "code_pipeline": "J18.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Bmr",
+      "code_pipeline": "M71.9",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/122_23070126__122_23070126_fusionne_cim10.json

@@ -0,0 +1,78 @@
+{
+  "dossier_id": "122_23070126/122_23070126_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Céphalées",
+    "code_pipeline": "G44.2",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Hypertension in",
+      "code_pipeline": "I15.9",
+      "confidence": "medium",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Hypertension intracrânienne",
+      "code_pipeline": "G93.2",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "N19",
+      "confidence": "medium",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Douleur : douleur",
+      "code_pipeline": "R52.2",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Désorientation",
+      "code_pipeline": "R41.0",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/125_23074494__125_23074494_fusionne_cim10.json

@@ -0,0 +1,98 @@
+{
+  "dossier_id": "125_23074494/125_23074494_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Fracture pertrochantérienne",
+    "code_pipeline": "S72.1",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "N19",
+      "confidence": "medium",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Anémie",
+      "code_pipeline": "D64.9",
+      "confidence": "medium",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Fracture fermée d'",
+      "code_pipeline": "T14.2",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Ostéoporose",
+      "code_pipeline": "M81.8",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Palpation de la hanche",
+      "code_pipeline": "S70",
+      "confidence": "medium",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Malaise",
+      "code_pipeline": "R45.2",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Fracture du col",
+      "code_pipeline": "S72.0",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Fracture fermée d'autres vertèbres cervicales précisées",
+      "code_pipeline": "S12.2",
+      "confidence": "high",
+      "source": "trackare",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/132_23080179__132_23080179_fusionne_cim10.json

@@ -0,0 +1,278 @@
+{
+  "dossier_id": "132_23080179/132_23080179_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Adénopathie",
+    "code_pipeline": "R59.9",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Dyslipidémie",
+      "code_pipeline": "E78.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Dyspnée",
+      "code_pipeline": "R06.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Autres atteintes",
+      "code_pipeline": "T06",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Lymphome diffus à grandes",
+      "code_pipeline": "C83.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Ostéolyse",
+      "code_pipeline": "M89.5",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Cervicale",
+      "code_pipeline": "M54.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Syndrome de",
+      "code_pipeline": "Q91.3",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Troubles du rythme supra-ventriculaire",
+      "code_pipeline": "I47.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Phlébites à répétition des membres inférieurs avec maladie post-phlébitique",
+      "code_pipeline": "I87.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Asthénie post-COVID-19",
+      "code_pipeline": "G93.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Tassement vertébral",
+      "code_pipeline": "M48.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Obésité (IMC 31.733)",
+      "code_pipeline": "E66.04",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Dysphagie",
+      "code_pipeline": "R13",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Adénopathies",
+      "code_pipeline": "R59.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Hypokaliémie",
+      "code_pipeline": "E87.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 16,
+      "texte_original": "Nausées et vomissement",
+      "code_pipeline": "R11",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 17,
+      "texte_original": "Dorsalgie",
+      "code_pipeline": "M54.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 18,
+      "texte_original": "Céphalées",
+      "code_pipeline": "G44",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 19,
+      "texte_original": "Sciatique",
+      "code_pipeline": "M54.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 20,
+      "texte_original": "Chimiothérapie",
+      "code_pipeline": "Z51.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 21,
+      "texte_original": "Épanchement pleural",
+      "code_pipeline": "J90",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 22,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 23,
+      "texte_original": "Leucocytose",
+      "code_pipeline": "D72.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 24,
+      "texte_original": "Thrombocytose",
+      "code_pipeline": "D69.6",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 25,
+      "texte_original": "Ventilation ventilation",
+      "code_pipeline": "Z99.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/134_23050890__134_23050890_fusionne_cim10.json

@@ -0,0 +1,158 @@
+{
+  "dossier_id": "134_23050890/134_23050890_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Epispadia",
+    "code_pipeline": "Q54.2",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Hypota orthostatique",
+      "code_pipeline": "I95.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Examen général",
+      "code_pipeline": "Z00.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Cervicale",
+      "code_pipeline": "M54.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Obésité",
+      "code_pipeline": "E66.99",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Dépression, trouble du mental",
+      "code_pipeline": "F32.30",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Dysplasie de hanche",
+      "code_pipeline": "Q65.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Extrophie vésicale",
+      "code_pipeline": "Q64.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Dysraphie antérieure",
+      "code_pipeline": "Q71.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Épispadias",
+      "code_pipeline": "Q64.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Coxarthrose",
+      "code_pipeline": "M16",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Obésité (IMC 36.731)",
+      "code_pipeline": "E66.05",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Infection",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Ventilationventilationventilation",
+      "code_pipeline": "Z99.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/143_23096917__143_23096917_fusionne_cim10.json

@@ -0,0 +1,158 @@
+{
+  "dossier_id": "143_23096917/143_23096917_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Appendicite aigue",
+    "code_pipeline": "K35.8",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Péritonite",
+      "code_pipeline": "K65.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Infection à E.Coli",
+      "code_pipeline": "A04.4",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Epanchement intra-abdominal",
+      "code_pipeline": "R19.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Appendicite abcédée et suppurée",
+      "code_pipeline": "K35.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Éruption cutanée médicamenteuse",
+      "code_pipeline": "L24.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Appendicite aigüe",
+      "code_pipeline": "K35.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Pathologique",
+      "code_pipeline": "F07.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Ascite",
+      "code_pipeline": "R18",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "C fc 104",
+      "code_pipeline": "C10.4",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Céphalées",
+      "code_pipeline": "G44",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Insuffisance hépatique aiguë",
+      "code_pipeline": "K71.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Insuffisance rénale aiguë",
+      "code_pipeline": "N17.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/145_23057452__145_23057452_fusionne_cim10.json

@@ -0,0 +1,138 @@
+{
+  "dossier_id": "145_23057452/145_23057452_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Isolement",
+    "code_pipeline": "Z29.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Glaucome",
+      "code_pipeline": "H40",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Hydrocéphalie",
+      "code_pipeline": "G91.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Troubles de la",
+      "code_pipeline": "F43.2",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Cervicale",
+      "code_pipeline": "M54.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Goutte",
+      "code_pipeline": "M10.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Insuffisance rénale chronique",
+      "code_pipeline": "N18.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Diabète de type 2 non équilibré",
+      "code_pipeline": "E11.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Examen général",
+      "code_pipeline": "Z04",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Céphalées",
+      "code_pipeline": "G44",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Leucocytose",
+      "code_pipeline": "D72.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/153_23102610__153_23102610_fusionne_cim10.json

@@ -0,0 +1,188 @@
+{
+  "dossier_id": "153_23102610/153_23102610_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Cétonurie",
+    "code_pipeline": "R82.4",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Infection urinaire",
+      "code_pipeline": "N39.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Bactériurie due à Citrobacter koseri",
+      "code_pipeline": "A49.1",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "À 15",
+      "code_pipeline": "Z03.8",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Diabète de type 2",
+      "code_pipeline": "E11.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Dyslipidémie",
+      "code_pipeline": "E78.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Pathologique",
+      "code_pipeline": "F07.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Insuffisance rénale chronique stade 3a",
+      "code_pipeline": "N18.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Troubles de la fonction sexuelle, hypogonadisme masculin",
+      "code_pipeline": "E29.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "C 9.4",
+      "code_pipeline": "C94",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Effets indésirables",
+      "code_pipeline": "Y88.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Hernie ombilicale",
+      "code_pipeline": "K42.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Acanthosis nigricans",
+      "code_pipeline": "L83",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Adénopathie",
+      "code_pipeline": "R59.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Perte de poids a",
+      "code_pipeline": "R63.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Insuffisance hépatique chronique",
+      "code_pipeline": "K72.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 16,
+      "texte_original": "Infection sévère",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/159_23107113__trackare-00270032-23107113_00270032_23107113_cim10.json

@@ -0,0 +1,248 @@
+{
+  "dossier_id": "159_23107113/trackare-00270032-23107113_00270032_23107113_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "100 100 100 100",
+    "code_pipeline": "Y90.4",
+    "confidence": "medium",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Éruption cutanée médicamenteuse",
+      "code_pipeline": "L24.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Embolie pulmonaire",
+      "code_pipeline": "I26.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Dénutrition",
+      "code_pipeline": "E43",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Pneumopathie",
+      "code_pipeline": "J18.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Toxoplasmose congénitale",
+      "code_pipeline": "P37.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Reflux gastro-oesophagien",
+      "code_pipeline": "K21.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Adénopathie",
+      "code_pipeline": "R59.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Rubéole",
+      "code_pipeline": "B06",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Fièvre q",
+      "code_pipeline": "A78",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Toxoplasmose",
+      "code_pipeline": "B58.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Neuropathique",
+      "code_pipeline": "R52.10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Rougeole",
+      "code_pipeline": "B05.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Bartonelle",
+      "code_pipeline": "A44.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Varicelle",
+      "code_pipeline": "B01.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 16,
+      "texte_original": "Pneumopathie varicelleuse",
+      "code_pipeline": "B01",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 17,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 18,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 19,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 20,
+      "texte_original": "Infection",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 21,
+      "texte_original": "Anémie ferriprive",
+      "code_pipeline": "D50",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 22,
+      "texte_original": "Insuffisance hépatique chronique",
+      "code_pipeline": "K72.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/160_23099448__160_23099448_fusionne_cim10.json

@@ -0,0 +1,128 @@
+{
+  "dossier_id": "160_23099448/160_23099448_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Isolement",
+    "code_pipeline": "Z29.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Lipodystrophie",
+      "code_pipeline": "E88.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Troubles du comportement alimentaire (restriction alimentaire, gouter sucré quotidien)",
+      "code_pipeline": "F50.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "C 6.3",
+      "code_pipeline": "C63",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Diabète de type 1",
+      "code_pipeline": "O24.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Urticaire",
+      "code_pipeline": "L50.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Polytraumatisme suite à un accident de moto",
+      "code_pipeline": "V29.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Allergie au pollen",
+      "code_pipeline": "J30.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/167_23104446__167_23104446_fusionne_cim10.json

@@ -0,0 +1,88 @@
+{
+  "dossier_id": "167_23104446/167_23104446_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Gynécologique",
+    "code_pipeline": "Z01.4",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Ascite",
+      "code_pipeline": "R18",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Convalescence",
+      "code_pipeline": "Z54.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Hématome",
+      "code_pipeline": "N85.7",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Constipation",
+      "code_pipeline": "K59.0",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "H / 220",
+      "code_pipeline": "N43",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Insuffisance rénale aiguë",
+      "code_pipeline": "N17.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Hta",
+      "code_pipeline": "I10",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/170_23077016__170_23077016_fusionne_cim10.json

@@ -0,0 +1,108 @@
+{
+  "dossier_id": "170_23077016/170_23077016_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "À 38",
+    "code_pipeline": "S38.10",
+    "confidence": "medium",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Retard excrétoire post-opératoire",
+      "code_pipeline": "R62",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Constipation",
+      "code_pipeline": "K59.0",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Malaise",
+      "code_pipeline": "R53",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Pâleur",
+      "code_pipeline": "R23.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Polyarthrite rhumatoïde",
+      "code_pipeline": "M06.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Infection",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Ventilation ventilation",
+      "code_pipeline": "Z99.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/173_23069373__173_23069373_fusionne_cim10.json

@@ -0,0 +1,128 @@
+{
+  "dossier_id": "173_23069373/173_23069373_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Goitre multinodulaire",
+    "code_pipeline": "E04.2",
+    "confidence": null,
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Infection urinaire",
+      "code_pipeline": "N39.0",
+      "confidence": "",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Dyspnée",
+      "code_pipeline": "R06.0",
+      "confidence": "",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Examen général",
+      "code_pipeline": "Z00.0",
+      "confidence": "",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Apnées du sommeil",
+      "code_pipeline": "G47.3",
+      "confidence": "",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Autre migraines",
+      "code_pipeline": "G43.8",
+      "confidence": "",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Cervicale",
+      "code_pipeline": "M54.2",
+      "confidence": "",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "À 08",
+      "code_pipeline": "A08",
+      "confidence": "",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Obésité (IMC 47.438)",
+      "code_pipeline": "E66.0",
+      "confidence": "",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Dysphonie",
+      "code_pipeline": "R49.0",
+      "confidence": "",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Dysphagie",
+      "code_pipeline": "R13",
+      "confidence": "",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Ventilationventilationventilation",
+      "code_pipeline": "R06.4",
+      "confidence": "",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/176_23124187__176_23124187_fusionne_cim10.json

@@ -0,0 +1,268 @@
+{
+  "dossier_id": "176_23124187/176_23124187_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Hta",
+    "code_pipeline": "I10",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Dyslipidémie",
+      "code_pipeline": "E78.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Pneumopathie",
+      "code_pipeline": "J18.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Tabagisme",
+      "code_pipeline": "T65.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Douleurs articulaires",
+      "code_pipeline": "M25.5",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Toux",
+      "code_pipeline": "R05",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Cardiopathie",
+      "code_pipeline": "I25.1",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Hypokaliémie",
+      "code_pipeline": "E87.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Arthralgie diffuse",
+      "code_pipeline": "M19.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Hippocratisme digital",
+      "code_pipeline": "R68.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Adénopathies cervicales",
+      "code_pipeline": "R73.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Insuffisance rénale chronique",
+      "code_pipeline": "N18.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Sepsis",
+      "code_pipeline": "R65.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Adénopathies",
+      "code_pipeline": "R59.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Agitation",
+      "code_pipeline": "R45.4",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 16,
+      "texte_original": "Syndrome de",
+      "code_pipeline": "Q91.3",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 17,
+      "texte_original": "Épanchement pleural",
+      "code_pipeline": "J90",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 18,
+      "texte_original": "Bronchectasies",
+      "code_pipeline": "J47",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 19,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 20,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 21,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 22,
+      "texte_original": "Méningite bactérienne",
+      "code_pipeline": "G00",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 23,
+      "texte_original": "Carcinome épidermoïde du poumon",
+      "code_pipeline": "C34.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 24,
+      "texte_original": "Hypercalcémie",
+      "code_pipeline": "E83.50",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/179_23126805__trackare-23005591-23126805_23005591_23126805_cim10.json

@@ -0,0 +1,138 @@
+{
+  "dossier_id": "179_23126805/trackare-23005591-23126805_23005591_23126805_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Allergies allergie",
+    "code_pipeline": "Z00.0",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Éruption cutanée médicamenteuse",
+      "code_pipeline": "L24.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Obésité (IMC 32.583)",
+      "code_pipeline": "E66.05",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Allergie à la pénicilline",
+      "code_pipeline": "Z88.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Uriticaire",
+      "code_pipeline": "L50",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "U du 12",
+      "code_pipeline": "D51",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "U du 09",
+      "code_pipeline": "U09.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Infection",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Anémie ferriprive",
+      "code_pipeline": "D50",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Insuffisance rénale aiguë",
+      "code_pipeline": "N17.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/17_23100690__17_23100690_fusionne_cim10.json

@@ -0,0 +1,228 @@
+{
+  "dossier_id": "17_23100690/17_23100690_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Pancréatite aiguë",
+    "code_pipeline": "K85.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Dyslipidémie",
+      "code_pipeline": "E78.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Éthylisme",
+      "code_pipeline": "F10.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Douleurs neuropathiques",
+      "code_pipeline": "R52.10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Pancréatite aigue",
+      "code_pipeline": "K85.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Transit",
+      "code_pipeline": "F95.0",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Diabète",
+      "code_pipeline": "E14",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Cystites récidivantes",
+      "code_pipeline": "N30.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Plastie mammaire",
+      "code_pipeline": "Z42.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Lithiase vésiculaire",
+      "code_pipeline": "K80.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Angiocholite",
+      "code_pipeline": "K83.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Obésité (IMC 30.302)",
+      "code_pipeline": "E66.94",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Cholécystite",
+      "code_pipeline": "K80.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Hématome",
+      "code_pipeline": "N85.7",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Céphalées",
+      "code_pipeline": "G44",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 16,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 17,
+      "texte_original": "Adhérences péri",
+      "code_pipeline": "K66.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 18,
+      "texte_original": "Infection",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 19,
+      "texte_original": "Iléus",
+      "code_pipeline": "K56.7",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 20,
+      "texte_original": "Pancréatite aiguë, sans précision",
+      "code_pipeline": "K85.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/183_23087212__183_23087212_fusionne_cim10.json

@@ -0,0 +1,88 @@
+{
+  "dossier_id": "183_23087212/183_23087212_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Ventilationventilation",
+    "code_pipeline": "Z99.1",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Infection urinaire",
+      "code_pipeline": "N39.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Bricker fragile et endommagé",
+      "code_pipeline": "Q61.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Obésité (IMC 31.231)",
+      "code_pipeline": "E66.94",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Colostomie",
+      "code_pipeline": "Z93.3",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Pyélonéphrite aiguë droite",
+      "code_pipeline": "N10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/186_23105969__186_23105969_fusionne_cim10.json

@@ -0,0 +1,128 @@
+{
+  "dossier_id": "186_23105969/186_23105969_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Prostatite aig",
+    "code_pipeline": "N41.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Fibrillation auriculaire",
+      "code_pipeline": "I48.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Dyslipidémie",
+      "code_pipeline": "E78.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Hypercholestérolémie",
+      "code_pipeline": "E78.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Cervicale",
+      "code_pipeline": "M54.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Trouble de la coagulation (TP abaissé)",
+      "code_pipeline": "D68.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Réaction transfusionnelle (RAI positive pour anti KEL 1)",
+      "code_pipeline": "T80.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Examen général",
+      "code_pipeline": "Z04",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Rétention urinaire aiguë",
+      "code_pipeline": "N13.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Hypertrophie bénigne de la prostate",
+      "code_pipeline": "N40",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Sur activité",
+      "code_pipeline": "Z53.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Infection",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/187_23133268__187_23133268_fusionne_cim10.json

@@ -0,0 +1,228 @@
+{
+  "dossier_id": "187_23133268/187_23133268_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Épisclerite",
+    "code_pipeline": "H15.1",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Dysphagie",
+      "code_pipeline": "R13",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Douleur au niveau de",
+      "code_pipeline": "R52.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Toux",
+      "code_pipeline": "R05",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Thrombose veineuse",
+      "code_pipeline": "I82.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Céphalées",
+      "code_pipeline": "G44",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Douleurs articulaires",
+      "code_pipeline": "M25.5",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Maladie de",
+      "code_pipeline": "Z22.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Gynécologique",
+      "code_pipeline": "Z01.4",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Dysthyroïdie",
+      "code_pipeline": "E06.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Aphtose récurrente",
+      "code_pipeline": "K12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Fièvre",
+      "code_pipeline": "R50.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Perte de poids",
+      "code_pipeline": "R63.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Obésité (IMC 31.616)",
+      "code_pipeline": "E66.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Épisclérites",
+      "code_pipeline": "H15.1",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Hépatite b",
+      "code_pipeline": "B16.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 16,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 17,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 18,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 19,
+      "texte_original": "Hépatite c",
+      "code_pipeline": "B17.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 20,
+      "texte_original": "Douleur abdominale",
+      "code_pipeline": "R10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/193_23123388__193_23123388_fusionne_cim10.json

@@ -0,0 +1,98 @@
+{
+  "dossier_id": "193_23123388/193_23123388_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Ventilationventilation",
+    "code_pipeline": "Z99.1",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Paraparésie spastique",
+      "code_pipeline": "G82.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Evolution postopératoire",
+      "code_pipeline": "Z54.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Sclérose en plaques",
+      "code_pipeline": "G35",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Céphalées",
+      "code_pipeline": "G44",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Hématome",
+      "code_pipeline": "N85.7",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Pneumothorax",
+      "code_pipeline": "J93.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Bmr",
+      "code_pipeline": "U83.71",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Infection",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/208_23151988__208_23151988_fusionne_cim10.json

@@ -0,0 +1,148 @@
+{
+  "dossier_id": "208_23151988/208_23151988_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Cholécystite aigue",
+    "code_pipeline": "K81.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Cholécystite aiguë",
+      "code_pipeline": "K81.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Adhérences péri",
+      "code_pipeline": "K66.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Calculs biliaires intra-vésiculaires",
+      "code_pipeline": "K80.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Inflammation du lit vésiculaire",
+      "code_pipeline": "K81.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Thrombose veineuse profonde",
+      "code_pipeline": "I80.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Ulcère d'estomac",
+      "code_pipeline": "K25",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Cholécystite, sans précision",
+      "code_pipeline": "K81.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Cholécystite",
+      "code_pipeline": "K80.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Hydrocholécyste",
+      "code_pipeline": "K82.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Hémorragie",
+      "code_pipeline": "I62.1",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/213_23156193__trackare-11010227-23156193_11010227_23156193_cim10.json

@@ -0,0 +1,48 @@
+{
+  "dossier_id": "213_23156193/trackare-11010227-23156193_11010227_23156193_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Colique néphrétique, sans précision",
+    "code_pipeline": "N23",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Et gynécologique",
+      "code_pipeline": "Z01.4",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Infection",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Allergie à la pénicilline",
+      "code_pipeline": "Z88.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/218_23164383__trackare-23022580-23164383_23022580_23164383_cim10.json

@@ -0,0 +1,118 @@
+{
+  "dossier_id": "218_23164383/trackare-23022580-23164383_23022580_23164383_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Neuronite (névrite) vestibulaire",
+    "code_pipeline": "H81.2",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Malaise",
+      "code_pipeline": "R53",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Asthme",
+      "code_pipeline": "J45.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Migraine",
+      "code_pipeline": "G43.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Céphalées",
+      "code_pipeline": "G44",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Cervicalgie",
+      "code_pipeline": "M54.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Névrite vestibulaire",
+      "code_pipeline": "H81.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Nausées et des vomissements",
+      "code_pipeline": "R11",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Système vertébro-basilaire",
+      "code_pipeline": "G45.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Infection",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/21_23111304__21_23111304_fusionne_cim10.json

@@ -0,0 +1,168 @@
+{
+  "dossier_id": "21_23111304/21_23111304_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Pancreatite aigue lithiasique",
+    "code_pipeline": "K85.1",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Infection urinaire",
+      "code_pipeline": "N39.0",
+      "confidence": "high",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Cystite",
+      "code_pipeline": "N30.9",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Bactériémie à Escherichia coli",
+      "code_pipeline": "B96.2",
+      "confidence": "high",
+      "source": "llm_das",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "À 09",
+      "code_pipeline": "A41.9",
+      "confidence": "medium",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Lithiase vésiculaire",
+      "code_pipeline": "K80.2",
+      "confidence": "high",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Cholangite",
+      "code_pipeline": "K81.0",
+      "confidence": "high",
+      "source": "llm_das",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Distension vésiculaire significative",
+      "code_pipeline": "K80.8",
+      "confidence": "medium",
+      "source": "llm_das",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Lésion hépatique gauche",
+      "code_pipeline": "K71.8",
+      "confidence": "medium",
+      "source": "llm_das",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Adhérences péri",
+      "code_pipeline": "K66.0",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Cholécystite lithiasique",
+      "code_pipeline": "K80.1",
+      "confidence": "high",
+      "source": "llm_das",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Pancreatite aigue",
+      "code_pipeline": "K85.0",
+      "confidence": "high",
+      "source": "regex",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Prophylactique",
+      "code_pipeline": "Z29.9",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Goute",
+      "code_pipeline": "M10.9",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Hernie ombilicale",
+      "code_pipeline": "K42.9",
+      "confidence": "high",
+      "source": "edsnlp",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Pancréatite aiguë, sans précision",
+      "code_pipeline": "K85.9",
+      "confidence": "high",
+      "source": "trackare",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/221_23167859__trackare-23023011-23167859_23023011_23167859_cim10.json

@@ -0,0 +1,158 @@
+{
+  "dossier_id": "221_23167859/trackare-23023011-23167859_23023011_23167859_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Hémorragie sous-durale non traumatique",
+    "code_pipeline": "I62.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Obésité (IMC 30.062)",
+      "code_pipeline": "E66.94",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Céphalée",
+      "code_pipeline": "G44",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Goutte",
+      "code_pipeline": "M10.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Colique néphrétique",
+      "code_pipeline": "N23",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Hématome",
+      "code_pipeline": "N85.7",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Épilepsie",
+      "code_pipeline": "G40.8",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Ataxie cérébelleuse",
+      "code_pipeline": "G11.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Surveillance de la surveillance",
+      "code_pipeline": "Z74.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Insuffisance rénale chronique",
+      "code_pipeline": "N18.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Hyperthermie",
+      "code_pipeline": "R50.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Faiblesse des membres / déficit",
+      "code_pipeline": "R20.8",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/223_23169043__223_23169043_fusionne_cim10.json

@@ -0,0 +1,218 @@
+{
+  "dossier_id": "223_23169043/223_23169043_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Pancreatite aigue",
+    "code_pipeline": "K85.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Hépatocellulaire",
+      "code_pipeline": "K76.8",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Hépatite aig",
+      "code_pipeline": "B17.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Pancréatite aiguë, sans précision",
+      "code_pipeline": "K85.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Incontinence urinaire",
+      "code_pipeline": "R32",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Diabète type",
+      "code_pipeline": "O24.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Pancréatique chronique",
+      "code_pipeline": "K86.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Hypertrophie de la",
+      "code_pipeline": "N90.6",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Cirrhose hépatique",
+      "code_pipeline": "K74.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Arthrose diffuse",
+      "code_pipeline": "M15.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Syndrome inflammatoire",
+      "code_pipeline": "R65.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Diabète de type 2",
+      "code_pipeline": "E11.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Pancréatite chronique",
+      "code_pipeline": "K86.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Diabète type 2",
+      "code_pipeline": "E11.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Céphalées",
+      "code_pipeline": "G44",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 16,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 17,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 18,
+      "texte_original": "Infection urinaire",
+      "code_pipeline": "N39.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 19,
+      "texte_original": "Insuffisance rénale aiguë",
+      "code_pipeline": "N17.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/225_23160703__225_23160703_fusionne_cim10.json

@@ -0,0 +1,68 @@
+{
+  "dossier_id": "225_23160703/225_23160703_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Ventilationventilation",
+    "code_pipeline": "Z99.1",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Adhérences abdominales",
+      "code_pipeline": "K56.3",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Globe vésical",
+      "code_pipeline": "C67.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Hématome",
+      "code_pipeline": "N85.7",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Leucocytose",
+      "code_pipeline": "D72.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/226_23175167__226_23175167_fusionne_cim10.json

@@ -0,0 +1,128 @@
+{
+  "dossier_id": "226_23175167/226_23175167_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Embolie et thrombose des artères des membres inférieurs",
+    "code_pipeline": "I74.3",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Antécédent de thrombose du pontage fémoro-distal",
+      "code_pipeline": "T82.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "X 20",
+      "code_pipeline": "X20",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Embolie et thrombose",
+      "code_pipeline": "I74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Hématome",
+      "code_pipeline": "N85.7",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Kyste rein",
+      "code_pipeline": "N28.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Anesthésie cutanée",
+      "code_pipeline": "R20.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Hypoesthésie",
+      "code_pipeline": "R20.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Goutte",
+      "code_pipeline": "M10.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Insuffisance rénale aiguë",
+      "code_pipeline": "N17.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Infection",
+      "code_pipeline": "A49.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/228_23176885__228_23176885_fusionne_cim10.json

@@ -0,0 +1,208 @@
+{
+  "dossier_id": "228_23176885/228_23176885_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Diabète de type 2",
+    "code_pipeline": "E11.8",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Infection urinaire",
+      "code_pipeline": "N39.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Agalactiae",
+      "code_pipeline": "O92.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Infection urinaire à Streptococcus agalactiae",
+      "code_pipeline": "N39.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "À 09",
+      "code_pipeline": "K09.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Diabète de type 2",
+      "code_pipeline": "E11.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Lésion de l'épaule",
+      "code_pipeline": "M75.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Ldl 0.4",
+      "code_pipeline": "L43.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Dyslipidémie",
+      "code_pipeline": "E78.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Obésité",
+      "code_pipeline": "E66.99",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Cancer du pancréas (antécédent familial)",
+      "code_pipeline": "C25.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Leucémie (antécédent familial)",
+      "code_pipeline": "Z80.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Thérapeutiques pour thérapeutiques",
+      "code_pipeline": "Z41.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 16,
+      "texte_original": "Bmr",
+      "code_pipeline": "U83.71",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 17,
+      "texte_original": "Bursite",
+      "code_pipeline": "M71.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 18,
+      "texte_original": "Cétonurie",
+      "code_pipeline": "R82.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/230_23167769__230_23167769_fusionne_cim10.json

@@ -0,0 +1,178 @@
+{
+  "dossier_id": "230_23167769/230_23167769_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Dyspnée",
+    "code_pipeline": "R06.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Pneumopathie",
+      "code_pipeline": "J18.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Toux",
+      "code_pipeline": "R05",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Examen général",
+      "code_pipeline": "Z00.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Asthme",
+      "code_pipeline": "J45.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Maladie de",
+      "code_pipeline": "Z22.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Cervicale",
+      "code_pipeline": "M54.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Insuffisance rénale chronique sévère mixte",
+      "code_pipeline": "N18.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Insuffisance cardiaque",
+      "code_pipeline": "I11.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Éventration",
+      "code_pipeline": "K42.0",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Insuffisance respiratoire",
+      "code_pipeline": "J96.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Colostomie",
+      "code_pipeline": "Z43.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Goutte",
+      "code_pipeline": "M10.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Leucocytose",
+      "code_pipeline": "D72.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Allergies allergie",
+      "code_pipeline": "Z00.0",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/234_23174515__234_23174515_fusionne_cim10.json

@@ -0,0 +1,98 @@
+{
+  "dossier_id": "234_23174515/234_23174515_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Épilepsie",
+    "code_pipeline": "G40.9",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Pathologique",
+      "code_pipeline": "F07.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Gliome",
+      "code_pipeline": "C70.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Tabagisme",
+      "code_pipeline": "T65.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Céphalées",
+      "code_pipeline": "G44",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Leucocytose",
+      "code_pipeline": "D72.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Risque de déséquilibre de l'épilepsie",
+      "code_pipeline": "G41.8",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Isolement",
+      "code_pipeline": "Z29.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/236_23116794__236_23116794_fusionne_cim10.json

@@ -0,0 +1,138 @@
+{
+  "dossier_id": "236_23116794/236_23116794_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Fistule colo-vaginale",
+    "code_pipeline": "N82.3",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Infection urinaire",
+      "code_pipeline": "N39.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "À 15",
+      "code_pipeline": "Z03.8",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Sigmoïdite",
+      "code_pipeline": "K51.9",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Diverticule pan colique",
+      "code_pipeline": "K51.8",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Obésité (IMC 32.258)",
+      "code_pipeline": "E66.04",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Fistule",
+      "code_pipeline": "N82.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Convalescence",
+      "code_pipeline": "Z54.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Nausées et vomissements",
+      "code_pipeline": "R11",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Goutte",
+      "code_pipeline": "M10.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Insuffisance cardiaque",
+      "code_pipeline": "I11.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Insuffisance rénale aiguë",
+      "code_pipeline": "N17.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Ventilation ventilation ventilation",
+      "code_pipeline": "Z99.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/240_23171519__240_23171519_fusionne_cim10.json

@@ -0,0 +1,178 @@
+{
+  "dossier_id": "240_23171519/240_23171519_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Chimiothérapie",
+    "code_pipeline": "Z51.1",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Adénocarcinome sigmoïdien métastatique hépatique",
+      "code_pipeline": "C78.7",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Embolisation portale",
+      "code_pipeline": "K92.8",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "F 23",
+      "code_pipeline": "F23",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Éruption cutanée médicamenteuse",
+      "code_pipeline": "L24.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Hypokaliémie",
+      "code_pipeline": "E87.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Syndrome main-pied",
+      "code_pipeline": "M71.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Alopécie chimio-induite",
+      "code_pipeline": "C76.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Hépatalgies",
+      "code_pipeline": "K71.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "N 2.5",
+      "code_pipeline": "N25",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Hépatite",
+      "code_pipeline": "K70",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Ventilationventilationventilation",
+      "code_pipeline": "Z99.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/245_23193113__trackare-03017963-23193113_03017963_23193113_cim10.json

@@ -0,0 +1,128 @@
+{
+  "dossier_id": "245_23193113/trackare-03017963-23193113_03017963_23193113_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Prostatite aiguë (infection urinaire masculine)",
+    "code_pipeline": "N41.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Infection urinaire",
+      "code_pipeline": "N39.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Sepsis",
+      "code_pipeline": "R65.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Obésité (IMC 31.02)",
+      "code_pipeline": "E66.99",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Prostatite aiguë",
+      "code_pipeline": "N41.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Colique néphrétique",
+      "code_pipeline": "N23",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Hypotension",
+      "code_pipeline": "I95.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Diverticulite",
+      "code_pipeline": "K57.3",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/246_23193699__246_23193699_fusionne_cim10.json

@@ -0,0 +1,88 @@
+{
+  "dossier_id": "246_23193699/246_23193699_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Fractures ouvertes d'autres parties de la jambe",
+    "code_pipeline": "S82.8",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Fracture du",
+      "code_pipeline": "S72.8",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Anesthésie : anesthésie",
+      "code_pipeline": "T41.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Infection post-opératoire de la fracture ouverte",
+      "code_pipeline": "T84.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Fracture ouverte",
+      "code_pipeline": "T14.2",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Fracture ouverte de",
+      "code_pipeline": "S22.31",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Fractures ouvertes d'autres",
+      "code_pipeline": "S92.21",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/37_23158940__37_23158940_fusionne_cim10.json

@@ -0,0 +1,118 @@
+{
+  "dossier_id": "37_23158940/37_23158940_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Cholécystite aiguë",
+    "code_pipeline": "K81.0",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Cholécystite aiguë",
+      "code_pipeline": "K81.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Calcul vésiculaire",
+      "code_pipeline": "K80.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Polype vésiculaire",
+      "code_pipeline": "N84.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Angiocholite",
+      "code_pipeline": "K83.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Insuffisance rénale aigue",
+      "code_pipeline": "N17.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Hépatite astreignante",
+      "code_pipeline": "K71.5",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Calcul des canaux biliaires (sans angiocholite ni cholécystite)",
+      "code_pipeline": "K80.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/38_23162619__trackare-13029354-23162619_13029354_23162619_cim10.json

@@ -0,0 +1,188 @@
+{
+  "dossier_id": "38_23162619/trackare-13029354-23162619_13029354_23162619_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Cholécystite, sans précision",
+    "code_pipeline": "K81.9",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Cholécystite aiguë",
+      "code_pipeline": "K81.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Hypertension artérielle",
+      "code_pipeline": "I10",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Sepsis",
+      "code_pipeline": "R65.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Obésité (IMC 33.659)",
+      "code_pipeline": "E66.05",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Cholécystite, sans précision",
+      "code_pipeline": "K81.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Cholécystite aigue",
+      "code_pipeline": "K80.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Cholécystite",
+      "code_pipeline": "K80.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Stimulateur",
+      "code_pipeline": "Y50.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Douleur aiguë",
+      "code_pipeline": "R52.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Méningite bactérienne",
+      "code_pipeline": "G00",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Sur activité",
+      "code_pipeline": "Z53.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Adhérences péri",
+      "code_pipeline": "K66.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Insuffisance rénale aiguë",
+      "code_pipeline": "N17.8",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 16,
+      "texte_original": "Infection post-opératoire",
+      "code_pipeline": "T84.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/54_23230165__54_23230165_fusionne_cim10.json

@@ -0,0 +1,178 @@
+{
+  "dossier_id": "54_23230165/54_23230165_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Cholecystite chronique",
+    "code_pipeline": "K81.1",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Ganglion du collet réactionnel",
+      "code_pipeline": "C77.8",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Pathologique",
+      "code_pipeline": "F07.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Cholélithiasis (épisode précédent)",
+      "code_pipeline": "K80.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Pancréatite aig",
+      "code_pipeline": "K85.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Lithiase vésiculaire",
+      "code_pipeline": "K80.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Tabagisme",
+      "code_pipeline": "T65.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Obésité (IMC 30.508)",
+      "code_pipeline": "E66.04",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Pancréatite aiguë",
+      "code_pipeline": "K85.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Plaie des voies biliaires",
+      "code_pipeline": "K83.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 10,
+      "texte_original": "Fibrose hépatique",
+      "code_pipeline": "K74.2",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 11,
+      "texte_original": "Thrombopénie",
+      "code_pipeline": "D69.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 12,
+      "texte_original": "Musculaire - masse musculaire",
+      "code_pipeline": "M62.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 13,
+      "texte_original": "Bmr",
+      "code_pipeline": "U83.71",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 14,
+      "texte_original": "Cholangite aiguë",
+      "code_pipeline": "K81.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 15,
+      "texte_original": "Abcès hépatique",
+      "code_pipeline": "K76.8",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/66_23001636__trackare-20018892-23001636_20018892_23001636_cim10.json

@@ -0,0 +1,88 @@
+{
+  "dossier_id": "66_23001636/trackare-20018892-23001636_20018892_23001636_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Ventilationventilationventilationventilationventilationventilationventilationventilation",
+    "code_pipeline": "Z99.1",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Dénutrition",
+      "code_pipeline": "E43",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Sepsis",
+      "code_pipeline": "R65.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Gastrostomie",
+      "code_pipeline": "Z93.1",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Cachexie",
+      "code_pipeline": "R64",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Perte de poids +",
+      "code_pipeline": "R63.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Anémie ferriprive",
+      "code_pipeline": "D50",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/68_23046068__trackare--23046068__23046068_cim10.json

@@ -0,0 +1,88 @@
+{
+  "dossier_id": "68_23046068/trackare--23046068__23046068_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Bmr",
+    "code_pipeline": "M71.9",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Dénutrition",
+      "code_pipeline": "E43",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Infection urinaire",
+      "code_pipeline": "N39.0",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Blastose",
+      "code_pipeline": "B40.8",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Perte de poids",
+      "code_pipeline": "R63.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Effets indésirables",
+      "code_pipeline": "Y88.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Ankylosé",
+      "code_pipeline": "M48.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/73_23139637__trackare-05001391-23139637_05001391_23139637_cim10.json

@@ -0,0 +1,58 @@
+{
+  "dossier_id": "73_23139637/trackare-05001391-23139637_05001391_23139637_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "À\n19",
+    "code_pipeline": "D19.7",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Ventilation ventilationventilation",
+      "code_pipeline": "Z99.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Dyspnée",
+      "code_pipeline": "R06.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Hématome",
+      "code_pipeline": "N85.7",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Troubles anxieux",
+      "code_pipeline": "F41.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/78_23187785__trackare-03017203-23187785_03017203_23187785_cim10.json

@@ -0,0 +1,118 @@
+{
+  "dossier_id": "78_23187785/trackare-03017203-23187785_03017203_23187785_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Bmr",
+    "code_pipeline": "M71.9",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Obésité (IMC 34.651)",
+      "code_pipeline": "E66.95",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "À la 03",
+      "code_pipeline": "N30.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Sarm",
+      "code_pipeline": "U82.10",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Ascite",
+      "code_pipeline": "R18",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Soins palliatif",
+      "code_pipeline": "Z51.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Encephalopathie",
+      "code_pipeline": "C71.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Prurit",
+      "code_pipeline": "L29.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Cirrhose hépatique",
+      "code_pipeline": "K74.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Insuffisance hépato-ciliaire (CHC)",
+      "code_pipeline": "K76.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Ictère",
+      "code_pipeline": "K81.9",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/79_23187785__79_23187785_fusionne_cim10.json

@@ -0,0 +1,118 @@
+{
+  "dossier_id": "79_23187785/79_23187785_fusionne_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Bmr",
+    "code_pipeline": "M71.9",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Obésité (IMC 34.651)",
+      "code_pipeline": "E66.95",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "À la 03",
+      "code_pipeline": "N30.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Sarm",
+      "code_pipeline": "U82.10",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Ascite",
+      "code_pipeline": "R18",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Soins palliatif",
+      "code_pipeline": "Z51.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Encephalopathie",
+      "code_pipeline": "C71.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Prurit",
+      "code_pipeline": "L29.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Cirrhose hépatique",
+      "code_pipeline": "K74.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Insuffisance hépato-ciliaire (CHC)",
+      "code_pipeline": "K76.6",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 9,
+      "texte_original": "Ictère",
+      "code_pipeline": "K81.9",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/83_23187785__trackcare_82_23187785_cim10.json

@@ -0,0 +1,108 @@
+{
+  "dossier_id": "83_23187785/trackcare_82_23187785_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Bmr",
+    "code_pipeline": "M71.9",
+    "confidence": "low",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Obésité (IMC 34.651)",
+      "code_pipeline": "E66.95",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "À la 03",
+      "code_pipeline": "N30.3",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Sarm",
+      "code_pipeline": "U82.10",
+      "confidence": "low",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Ascite",
+      "code_pipeline": "R18",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Soins palliatif",
+      "code_pipeline": "Z51.5",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Encephalopathie",
+      "code_pipeline": "C71.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 6,
+      "texte_original": "Prurit",
+      "code_pipeline": "L29.9",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 7,
+      "texte_original": "Cirrhose",
+      "code_pipeline": "K74",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 8,
+      "texte_original": "Insuffisance hépatique chronique (CHC)",
+      "code_pipeline": "K72.1",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/84_23215994__trackare-16014215-23215994_16014215_23215994_cim10.json

@@ -0,0 +1,78 @@
+{
+  "dossier_id": "84_23215994/trackare-16014215-23215994_16014215_23215994_cim10",
+  "validateur": "",
+  "date_validation": "",
+  "statut": "non_commence",
+  "dp": {
+    "texte_original": "Urticaire",
+    "code_pipeline": "L50.1",
+    "confidence": "high",
+    "statut": "correct",
+    "code_corrige": null,
+    "commentaire": ""
+  },
+  "das": [
+    {
+      "index": 0,
+      "texte_original": "Insuffisance rénale",
+      "code_pipeline": "I12.0",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 1,
+      "texte_original": "Anémie",
+      "code_pipeline": "D55.9",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 2,
+      "texte_original": "Anémie par carence",
+      "code_pipeline": "D50",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 3,
+      "texte_original": "Blastose",
+      "code_pipeline": "B40.8",
+      "confidence": "medium",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 4,
+      "texte_original": "Faux anévrisme huméral gauche",
+      "code_pipeline": "I72.4",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    },
+    {
+      "index": 5,
+      "texte_original": "Pose de VVP (Voie de Veinoportée)",
+      "code_pipeline": "Z95.80",
+      "confidence": "high",
+      "source": "",
+      "statut": "correct",
+      "code_corrige": null,
+      "commentaire": ""
+    }
+  ],
+  "das_ajoutes": [],
+  "commentaire_general": ""
+}

data/gold_standard/_selection.json

@@ -0,0 +1,58 @@
+{
+  "date_selection": "2026-02-17T21:35:32",
+  "total": 50,
+  "cpam": 25,
+  "non_cpam": 25,
+  "dossiers": [
+    "116_23065570/116_23065570_fusionne_cim10",
+    "132_23080179/132_23080179_fusionne_cim10",
+    "134_23050890/134_23050890_fusionne_cim10",
+    "143_23096917/143_23096917_fusionne_cim10",
+    "145_23057452/145_23057452_fusionne_cim10",
+    "153_23102610/153_23102610_fusionne_cim10",
+    "170_23077016/170_23077016_fusionne_cim10",
+    "176_23124187/176_23124187_fusionne_cim10",
+    "17_23100690/17_23100690_fusionne_cim10",
+    "183_23087212/183_23087212_fusionne_cim10",
+    "186_23105969/186_23105969_fusionne_cim10",
+    "193_23123388/193_23123388_fusionne_cim10",
+    "208_23151988/208_23151988_fusionne_cim10",
+    "21_23111304/21_23111304_fusionne_cim10",
+    "223_23169043/223_23169043_fusionne_cim10",
+    "225_23160703/225_23160703_fusionne_cim10",
+    "226_23175167/226_23175167_fusionne_cim10",
+    "228_23176885/228_23176885_fusionne_cim10",
+    "230_23167769/230_23167769_fusionne_cim10",
+    "234_23174515/234_23174515_fusionne_cim10",
+    "236_23116794/236_23116794_fusionne_cim10",
+    "240_23171519/240_23171519_fusionne_cim10",
+    "246_23193699/246_23193699_fusionne_cim10",
+    "37_23158940/37_23158940_fusionne_cim10",
+    "54_23230165/54_23230165_fusionne_cim10",
+    "122_23070126/122_23070126_fusionne_cim10",
+    "218_23164383/trackare-23022580-23164383_23022580_23164383_cim10",
+    "221_23167859/trackare-23023011-23167859_23023011_23167859_cim10",
+    "113_23065949/trackare-18022356-23065949_18022356_23065949_cim10",
+    "38_23162619/trackare-13029354-23162619_13029354_23162619_cim10",
+    "83_23187785/trackcare_82_23187785_cim10",
+    "84_23215994/trackare-16014215-23215994_16014215_23215994_cim10",
+    "173_23069373/173_23069373_fusionne_cim10",
+    "213_23156193/trackare-11010227-23156193_11010227_23156193_cim10",
+    "245_23193113/trackare-03017963-23193113_03017963_23193113_cim10",
+    "73_23139637/trackare-05001391-23139637_05001391_23139637_cim10",
+    "115_23066188/115_23066188_fusionne_cim10",
+    "103_23056749/103_23056749_fusionne_cim10",
+    "125_23074494/125_23074494_fusionne_cim10",
+    "66_23001636/trackare-20018892-23001636_20018892_23001636_cim10",
+    "159_23107113/trackare-00270032-23107113_00270032_23107113_cim10",
+    "160_23099448/160_23099448_fusionne_cim10",
+    "187_23133268/187_23133268_fusionne_cim10",
+    "179_23126805/trackare-23005591-23126805_23005591_23126805_cim10",
+    "118_23042633/118_23042633_fusionne_cim10",
+    "119_23062819/119_23062819_fusionne_cim10",
+    "78_23187785/trackare-03017203-23187785_03017203_23187785_cim10",
+    "79_23187785/79_23187785_fusionne_cim10",
+    "167_23104446/167_23104446_fusionne_cim10",
+    "68_23046068/trackare--23046068__23046068_cim10"
+  ]
+}

data/referentiels/069835fec62e_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/137b40623e82_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/1703dd92913f_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/194272ef6cb6_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/1a40d11c7ce2_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/1d2ca361391c_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/1d9c65844377_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/1fc3dd81974f_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/23803b25e747_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/26ef0ab4e147_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/2cc26c4814fa_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/34a82abfeed8_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/3a3b837171bd_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/3d66e17ba0d0_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/3decf1f3165d_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/3ea7e3a4ed9f_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/40ee4a3989a8_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/433426770a3f_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/4482c0366139_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/4cbf5a2f32e9_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/54e1b1dd4a67_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/56f212d01d3b_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/590372dfec4d_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/5d8a689f1382_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/5fd99d4c190e_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/64fc5d4c3823_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/6d25cb9c160a_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/76e115a64038_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/7a4733e0d5e4_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/7b1facd1c4be_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/7b25f0fd65b3_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/808d8b3e099d_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/84cb164bb8f5_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/8cb453139f03_doc.txt

@@ -0,0 +1 @@
+test content

data/referentiels/8f013169f274_doc.txt

@@ -0,0 +1 @@
+test content