feat: RAG CPAM dédié avec requêtes multi-ciblées + prompt 3 axes
- Nouvelle search_similar_cpam() : priorité Guide Méthodo, seuil 0.40, déduplication par code CIM-10, fetch élargi top_k*3 - _search_rag_for_control() refactoré : 2-3 requêtes ciblées (codes contestés, argument CPAM, contexte clinique) au lieu d'1 fourre-tout - Fusion dédupliquée par (document, code, page), top 10 résultats - Prompt CPAM enrichi : 3 axes (médical, asymétrie, réglementaire), section asymétrie d'information, format réponse structuré - 9 nouveaux tests unitaires pour la logique RAG multi-requêtes Élimine les sources CIM-10 hors-sujet (F45, F98.1, F50.5 sur pancréatite) au profit de résultats Guide Méthodo et référentiels pertinents. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
@@ -1,11 +1,26 @@
|
||||
"""Tests pour la génération de contre-argumentation CPAM."""
|
||||
|
||||
from unittest.mock import patch
|
||||
from unittest.mock import patch, call
|
||||
|
||||
import pytest
|
||||
|
||||
from src.config import ControleCPAM, Diagnostic, DossierMedical, RAGSource, Sejour
|
||||
from src.control.cpam_response import _build_cpam_prompt, _format_response, generate_cpam_response
|
||||
from src.config import (
|
||||
ActeCCAM,
|
||||
BiologieCle,
|
||||
ControleCPAM,
|
||||
Diagnostic,
|
||||
DossierMedical,
|
||||
Imagerie,
|
||||
RAGSource,
|
||||
Sejour,
|
||||
Traitement,
|
||||
)
|
||||
from src.control.cpam_response import (
|
||||
_build_cpam_prompt,
|
||||
_format_response,
|
||||
_search_rag_for_control,
|
||||
generate_cpam_response,
|
||||
)
|
||||
|
||||
|
||||
def _make_dossier() -> DossierMedical:
|
||||
@@ -24,6 +39,37 @@ def _make_dossier() -> DossierMedical:
|
||||
)
|
||||
|
||||
|
||||
def _make_dossier_complet() -> DossierMedical:
|
||||
"""Crée un dossier médical enrichi avec traitements, imagerie, antécédents."""
|
||||
return DossierMedical(
|
||||
source_file="test.pdf",
|
||||
document_type="crh",
|
||||
sejour=Sejour(sexe="M", age=65, duree_sejour=5, imc=31.2),
|
||||
diagnostic_principal=Diagnostic(
|
||||
texte="Cholécystite aiguë",
|
||||
cim10_suggestion="K81.0",
|
||||
),
|
||||
diagnostics_associes=[
|
||||
Diagnostic(texte="Iléus réflexe", cim10_suggestion="K56.0"),
|
||||
],
|
||||
actes_ccam=[
|
||||
ActeCCAM(texte="Cholécystectomie", code_ccam_suggestion="HMFC004"),
|
||||
],
|
||||
biologie_cle=[
|
||||
BiologieCle(test="CRP", valeur="180 mg/L", anomalie=True),
|
||||
BiologieCle(test="Créatinine", valeur="450 µmol/L", anomalie=True),
|
||||
],
|
||||
imagerie=[
|
||||
Imagerie(type="Scanner abdominal", conclusion="Lithiase cholédocienne confirmée"),
|
||||
],
|
||||
traitements_sortie=[
|
||||
Traitement(medicament="Augmentin IV", posologie="3g/j"),
|
||||
Traitement(medicament="Morphine SC"),
|
||||
],
|
||||
antecedents=["HTA", "Diabète type 2"],
|
||||
)
|
||||
|
||||
|
||||
def _make_controle() -> ControleCPAM:
|
||||
"""Crée un contrôle CPAM de test."""
|
||||
return ControleCPAM(
|
||||
@@ -75,32 +121,115 @@ class TestBuildPrompt:
|
||||
assert "CIM-10 FR 2026" in prompt
|
||||
assert "page 64" in prompt
|
||||
|
||||
def test_prompt_contains_three_axes(self):
|
||||
dossier = _make_dossier()
|
||||
controle = _make_controle()
|
||||
prompt = _build_cpam_prompt(dossier, controle, [])
|
||||
|
||||
assert "AXE MÉDICAL" in prompt
|
||||
assert "AXE ASYMÉTRIE D'INFORMATION" in prompt
|
||||
assert "AXE RÉGLEMENTAIRE" in prompt
|
||||
|
||||
def test_prompt_contains_traitements_imagerie_when_present(self):
|
||||
dossier = _make_dossier_complet()
|
||||
controle = _make_controle()
|
||||
prompt = _build_cpam_prompt(dossier, controle, [])
|
||||
|
||||
assert "Augmentin IV 3g/j" in prompt
|
||||
assert "Morphine SC" in prompt
|
||||
assert "Scanner abdominal" in prompt
|
||||
assert "Lithiase cholédocienne confirmée" in prompt
|
||||
assert "HTA" in prompt
|
||||
assert "Diabète type 2" in prompt
|
||||
assert "IMC : 31.2" in prompt
|
||||
|
||||
def test_prompt_asymetrie_section_when_data_present(self):
|
||||
dossier = _make_dossier_complet()
|
||||
controle = _make_controle()
|
||||
prompt = _build_cpam_prompt(dossier, controle, [])
|
||||
|
||||
assert "ÉLÉMENTS DU DOSSIER NON TRANSMIS À LA CPAM" in prompt
|
||||
assert "CRP: 180 mg/L (anormale)" in prompt
|
||||
assert "Cholécystectomie (HMFC004)" in prompt
|
||||
|
||||
def test_prompt_no_asymetrie_section_when_no_data(self):
|
||||
dossier = DossierMedical(
|
||||
source_file="test.pdf",
|
||||
document_type="crh",
|
||||
sejour=Sejour(),
|
||||
diagnostic_principal=Diagnostic(texte="Test", cim10_suggestion="Z00"),
|
||||
)
|
||||
controle = _make_controle()
|
||||
prompt = _build_cpam_prompt(dossier, controle, [])
|
||||
|
||||
assert "ÉLÉMENTS DU DOSSIER NON TRANSMIS À LA CPAM" not in prompt
|
||||
|
||||
def test_prompt_json_format_new_fields(self):
|
||||
dossier = _make_dossier()
|
||||
controle = _make_controle()
|
||||
prompt = _build_cpam_prompt(dossier, controle, [])
|
||||
|
||||
assert "contre_arguments_medicaux" in prompt
|
||||
assert "contre_arguments_asymetrie" in prompt
|
||||
assert "contre_arguments_reglementaires" in prompt
|
||||
|
||||
|
||||
class TestFormatResponse:
|
||||
def test_full_response(self):
|
||||
def test_full_response_new_format(self):
|
||||
parsed = {
|
||||
"analyse_contestation": "La CPAM conteste le DAS K56.0",
|
||||
"points_accord": "Aucun",
|
||||
"contre_arguments": "Le guide méthodologique précise...",
|
||||
"references": "Guide métho p.64",
|
||||
"contre_arguments_medicaux": "Le guide méthodologique précise...",
|
||||
"contre_arguments_asymetrie": "La biologie montre une CRP à 180...",
|
||||
"contre_arguments_reglementaires": "L'UCR interprète restrictivement...",
|
||||
"references": "Guide métho p.64, CIM-10 K56.0",
|
||||
"conclusion": "Le DAS est justifié",
|
||||
}
|
||||
text = _format_response(parsed)
|
||||
|
||||
assert "ANALYSE DE LA CONTESTATION" in text
|
||||
assert "CONTRE-ARGUMENTS" in text
|
||||
assert "CONTRE-ARGUMENTS MÉDICAUX" in text
|
||||
assert "ASYMÉTRIE D'INFORMATION" in text
|
||||
assert "CONTRE-ARGUMENTS RÉGLEMENTAIRES" in text
|
||||
assert "REFERENCES" in text
|
||||
assert "CONCLUSION" in text
|
||||
# "Aucun" ne doit pas générer la section points d'accord
|
||||
assert "POINTS D'ACCORD" not in text
|
||||
# L'ancien champ ne doit pas apparaître
|
||||
assert "CONTRE-ARGUMENTS\n" not in text
|
||||
|
||||
def test_partial_response(self):
|
||||
def test_fallback_old_format(self):
|
||||
"""L'ancien champ contre_arguments est toujours géré (réponses en cache)."""
|
||||
parsed = {
|
||||
"contre_arguments": "Arguments...",
|
||||
"analyse_contestation": "Analyse...",
|
||||
"contre_arguments": "Arguments anciens...",
|
||||
"conclusion": "Conclusion...",
|
||||
}
|
||||
text = _format_response(parsed)
|
||||
|
||||
assert "CONTRE-ARGUMENTS" in text
|
||||
assert "CONTRE-ARGUMENTS\nArguments anciens..." in text
|
||||
assert "CONCLUSION" in text
|
||||
|
||||
def test_new_fields_override_fallback(self):
|
||||
"""Si les nouveaux champs existent, l'ancien contre_arguments est ignoré."""
|
||||
parsed = {
|
||||
"contre_arguments_medicaux": "Médicaux...",
|
||||
"contre_arguments": "Ancien fallback...",
|
||||
"conclusion": "Conclusion...",
|
||||
}
|
||||
text = _format_response(parsed)
|
||||
|
||||
assert "CONTRE-ARGUMENTS MÉDICAUX" in text
|
||||
assert "Ancien fallback" not in text
|
||||
|
||||
def test_partial_response(self):
|
||||
parsed = {
|
||||
"contre_arguments_medicaux": "Arguments médicaux...",
|
||||
"conclusion": "Conclusion...",
|
||||
}
|
||||
text = _format_response(parsed)
|
||||
|
||||
assert "CONTRE-ARGUMENTS MÉDICAUX" in text
|
||||
assert "CONCLUSION" in text
|
||||
|
||||
def test_empty_response(self):
|
||||
@@ -117,7 +246,8 @@ class TestGenerateResponse:
|
||||
]
|
||||
mock_ollama.return_value = {
|
||||
"analyse_contestation": "Analyse...",
|
||||
"contre_arguments": "Contre-arguments...",
|
||||
"contre_arguments_medicaux": "Contre-arguments médicaux...",
|
||||
"contre_arguments_asymetrie": "Asymétrie...",
|
||||
"conclusion": "Conclusion...",
|
||||
}
|
||||
|
||||
@@ -126,7 +256,7 @@ class TestGenerateResponse:
|
||||
|
||||
text, sources = generate_cpam_response(dossier, controle)
|
||||
|
||||
assert "Contre-arguments..." in text
|
||||
assert "Contre-arguments médicaux..." in text
|
||||
assert len(sources) == 1
|
||||
assert sources[0].document == "guide_methodo"
|
||||
mock_ollama.assert_called_once()
|
||||
@@ -144,3 +274,165 @@ class TestGenerateResponse:
|
||||
|
||||
assert text == ""
|
||||
assert sources == []
|
||||
|
||||
|
||||
class TestSearchRagForControl:
|
||||
"""Tests pour la logique de recherche RAG multi-requêtes."""
|
||||
|
||||
@patch("src.medical.rag_search.search_similar_cpam")
|
||||
def test_multiple_queries_with_da_ucr(self, mock_search):
|
||||
"""Avec da_ucr, on doit avoir au moins 2 requêtes (codes + argument)."""
|
||||
mock_search.return_value = [
|
||||
{"document": "guide_methodo", "page": 10, "code": None, "score": 0.6,
|
||||
"extrait": "Texte guide"},
|
||||
]
|
||||
|
||||
dossier = _make_dossier()
|
||||
controle = _make_controle() # da_ucr="K56.0"
|
||||
|
||||
results = _search_rag_for_control(controle, dossier)
|
||||
|
||||
# Au moins 2 appels : codes contestés + argument CPAM
|
||||
assert mock_search.call_count >= 2
|
||||
|
||||
@patch("src.medical.rag_search.search_similar_cpam")
|
||||
def test_query_codes_contains_cma(self, mock_search):
|
||||
"""La requête codes contestés doit contenir 'CMA' pour un DA."""
|
||||
mock_search.return_value = []
|
||||
|
||||
dossier = _make_dossier()
|
||||
controle = _make_controle() # da_ucr="K56.0"
|
||||
|
||||
_search_rag_for_control(controle, dossier)
|
||||
|
||||
# Premier appel = requête codes
|
||||
first_call_query = mock_search.call_args_list[0][0][0]
|
||||
assert "K56.0" in first_call_query
|
||||
assert "CMA" in first_call_query
|
||||
|
||||
@patch("src.medical.rag_search.search_similar_cpam")
|
||||
def test_query_argument_contains_titre(self, mock_search):
|
||||
"""La requête argument doit contenir le titre du contrôle."""
|
||||
mock_search.return_value = []
|
||||
|
||||
dossier = _make_dossier()
|
||||
controle = _make_controle()
|
||||
|
||||
_search_rag_for_control(controle, dossier)
|
||||
|
||||
# Deuxième appel = requête argument
|
||||
second_call_query = mock_search.call_args_list[1][0][0]
|
||||
assert controle.titre in second_call_query
|
||||
|
||||
@patch("src.medical.rag_search.search_similar_cpam")
|
||||
def test_deduplication_by_document_code_page(self, mock_search):
|
||||
"""Les résultats dupliqués (même document/code/page) sont fusionnés."""
|
||||
# Les deux requêtes retournent le même résultat
|
||||
shared_result = {
|
||||
"document": "guide_methodo", "page": 64, "code": None,
|
||||
"score": 0.55, "extrait": "Texte partagé",
|
||||
}
|
||||
mock_search.return_value = [shared_result.copy()]
|
||||
|
||||
dossier = _make_dossier()
|
||||
controle = _make_controle()
|
||||
|
||||
results = _search_rag_for_control(controle, dossier)
|
||||
|
||||
# Le résultat ne doit apparaître qu'une seule fois malgré les requêtes multiples
|
||||
guide_results = [r for r in results if r["document"] == "guide_methodo" and r.get("page") == 64]
|
||||
assert len(guide_results) == 1
|
||||
|
||||
@patch("src.medical.rag_search.search_similar_cpam")
|
||||
def test_dedup_keeps_best_score(self, mock_search):
|
||||
"""La déduplication garde le meilleur score."""
|
||||
def side_effect(query, top_k=8):
|
||||
if "CMA" in query:
|
||||
return [{"document": "cim10", "code": "K56.0", "page": None,
|
||||
"score": 0.5, "extrait": "Iléus"}]
|
||||
else:
|
||||
return [{"document": "cim10", "code": "K56.0", "page": None,
|
||||
"score": 0.7, "extrait": "Iléus"}]
|
||||
mock_search.side_effect = side_effect
|
||||
|
||||
dossier = _make_dossier()
|
||||
controle = _make_controle()
|
||||
|
||||
results = _search_rag_for_control(controle, dossier)
|
||||
|
||||
k56_results = [r for r in results if r.get("code") == "K56.0"]
|
||||
assert len(k56_results) == 1
|
||||
assert k56_results[0]["score"] == 0.7
|
||||
|
||||
@patch("src.medical.rag_search.search_similar_cpam")
|
||||
def test_no_codes_contestes_only_argument_query(self, mock_search):
|
||||
"""Sans codes contestés, seule la requête argument est lancée."""
|
||||
mock_search.return_value = []
|
||||
|
||||
dossier = _make_dossier()
|
||||
controle = ControleCPAM(
|
||||
numero_ogc=1,
|
||||
titre="Désaccord sur la durée",
|
||||
arg_ucr="Séjour trop long selon l'UCR.",
|
||||
decision_ucr="Rejet",
|
||||
dp_ucr=None,
|
||||
da_ucr=None,
|
||||
)
|
||||
|
||||
_search_rag_for_control(controle, dossier)
|
||||
|
||||
# Un seul appel : requête argument (pas de codes contestés)
|
||||
assert mock_search.call_count == 1
|
||||
|
||||
@patch("src.medical.rag_search.search_similar_cpam")
|
||||
def test_dp_ucr_query_contains_diagnostic_principal(self, mock_search):
|
||||
"""Avec dp_ucr, la requête codes mentionne 'diagnostic principal'."""
|
||||
mock_search.return_value = []
|
||||
|
||||
dossier = _make_dossier()
|
||||
controle = ControleCPAM(
|
||||
numero_ogc=2,
|
||||
titre="Désaccord sur le DP",
|
||||
arg_ucr="Le DP devrait être K80.1",
|
||||
decision_ucr="Rejet",
|
||||
dp_ucr="K81.0",
|
||||
da_ucr=None,
|
||||
)
|
||||
|
||||
_search_rag_for_control(controle, dossier)
|
||||
|
||||
first_call_query = mock_search.call_args_list[0][0][0]
|
||||
assert "K81.0" in first_call_query
|
||||
assert "diagnostic principal" in first_call_query
|
||||
|
||||
@patch("src.medical.rag_search.search_similar_cpam")
|
||||
def test_max_10_results(self, mock_search):
|
||||
"""Le résultat final est limité à 10 entrées."""
|
||||
mock_search.return_value = [
|
||||
{"document": "guide_methodo", "page": i, "code": None,
|
||||
"score": 0.9 - i * 0.01, "extrait": f"Texte {i}"}
|
||||
for i in range(8)
|
||||
]
|
||||
|
||||
dossier = _make_dossier()
|
||||
controle = _make_controle()
|
||||
|
||||
results = _search_rag_for_control(controle, dossier)
|
||||
|
||||
assert len(results) <= 10
|
||||
|
||||
@patch("src.medical.rag_search.search_similar_cpam")
|
||||
def test_clinical_query_when_das_match(self, mock_search):
|
||||
"""Requête clinique lancée quand da_ucr matche un DAS du dossier."""
|
||||
mock_search.return_value = []
|
||||
|
||||
dossier = _make_dossier() # DAS K56.0 "Iléus réflexe"
|
||||
controle = _make_controle() # da_ucr="K56.0"
|
||||
|
||||
_search_rag_for_control(controle, dossier)
|
||||
|
||||
# 3 appels : codes + argument + clinique
|
||||
assert mock_search.call_count == 3
|
||||
third_call_query = mock_search.call_args_list[2][0][0]
|
||||
assert "Iléus réflexe" in third_call_query
|
||||
assert "Cholécystite aiguë" in third_call_query
|
||||
|
||||
Reference in New Issue
Block a user