chore: mise à jour output pipeline (anonymized + structured)

Résultats de re-traitement pipeline v2 sur 261 dossiers.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
dom
2026-03-07 23:14:42 +01:00
parent c73515ac89
commit 13fe9fa666
734 changed files with 157158 additions and 304963 deletions

View File

@@ -0,0 +1,39 @@
{
"source_file": "CONSULTATION ANESTHESISTE 23098838.pdf",
"total_replacements": 37,
"regex_replacements": 24,
"ner_replacements": 1,
"sweep_replacements": 12,
"entities_found": [
{
"original": "23097921",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23097921",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "226 RUE DE L ETANG N° Tél : [TEL_1]",
"replacement": "[ADRESSE_2]",
"source": "regex",
"category": "adresse"
},
{
"original": "226 RUE DE L ETANG N° Tél : [TEL_1]",
"replacement": "[ADRESSE_2]",
"source": "regex",
"category": "adresse"
},
{
"original": "Mallampati",
"replacement": "[PERSONNE_1]",
"source": "ner",
"score": 0.8441555500030518
}
]
}

View File

@@ -0,0 +1,27 @@
{
"source_file": "CRO 23098838.pdf",
"total_replacements": 13,
"regex_replacements": 10,
"ner_replacements": 3,
"sweep_replacements": 0,
"entities_found": [
{
"original": "Dufour",
"replacement": "[PERSONNE_1]",
"source": "ner",
"score": 0.9303383827209473
},
{
"original": "MICHEL",
"replacement": "[PATIENT_1]",
"source": "ner",
"score": 0.9149897694587708
},
{
"original": "PRIOT",
"replacement": "[PATIENT_1]",
"source": "ner",
"score": 0.9843079447746277
}
]
}

View File

@@ -0,0 +1,597 @@
{
"source_file": "trackare-14002571-23098838_14002571_23098838.pdf",
"total_replacements": 718,
"regex_replacements": 256,
"ner_replacements": 54,
"sweep_replacements": 408,
"entities_found": [
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "14002571",
"replacement": "[IPP_1]",
"source": "regex",
"category": "ipp"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "23098838",
"replacement": "[EPISODE_1]",
"source": "regex",
"category": "episode"
},
{
"original": "640000162",
"replacement": "[FINESS]",
"source": "regex",
"category": "finess"
},
{
"original": "PARIS 13",
"replacement": "[LIEU_NAISS_1]",
"source": "regex",
"category": "lieu_naissance"
},
{
"original": "CHIRURGIE UROLOGIE - CHIRURGIE UROLOGIE\nDossier Patient\nDétails des patients",
"replacement": "[PATIENT_2]",
"source": "regex",
"category": "patient"
},
{
"original": "226 RUE DE L ETANG Ville de résidence: ONDRES",
"replacement": "[ADRESSE_1]",
"source": "regex",
"category": "adresse"
},
{
"original": "BURTIN",
"replacement": "[PERSONNE_1]",
"source": "ner",
"score": 0.8693852424621582
},
{
"original": "CURUTCHET",
"replacement": "[MEDECIN_7]",
"source": "ner",
"score": 0.8730018734931946
},
{
"original": "Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9977842569351196
},
{
"original": "Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9979665875434875
},
{
"original": "VERGEZ Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9632388353347778
},
{
"original": "VERGEZ Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.927723228931427
},
{
"original": "VERGEZ Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.8395925164222717
},
{
"original": "VERGEZ Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.8410494923591614
},
{
"original": "GLYCEMIE CAPILLAIRE",
"replacement": "[PERSONNE_2]",
"source": "ner",
"score": 0.9952131509780884
},
{
"original": "LYCEMIE CAPILLAIRE",
"replacement": "[PERSONNE_2]",
"source": "ner",
"score": 0.9931692481040955
},
{
"original": "GLYCEMIE CAPILLAIRE",
"replacement": "[PERSONNE_2]",
"source": "ner",
"score": 0.9615988731384277
},
{
"original": "GLYCEMIE CAPILLAIRE",
"replacement": "[PERSONNE_2]",
"source": "ner",
"score": 0.9621697068214417
},
{
"original": "Jeanne Signé",
"replacement": "[PERSONNE_3]",
"source": "ner",
"score": 0.9903035759925842
},
{
"original": "Jeanne Signé",
"replacement": "[PERSONNE_3]",
"source": "ner",
"score": 0.9867646098136902
},
{
"original": "DECLERCK Marie- Jeanne Signé",
"replacement": "[SOIGNANT_7]",
"source": "ner",
"score": 0.9395685195922852
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.846271812915802
},
{
"original": "ENTRESTO",
"replacement": "[PERSONNE_4]",
"source": "ner",
"score": 0.9336079359054565
},
{
"original": "ENTRESTO",
"replacement": "[PERSONNE_4]",
"source": "ner",
"score": 0.8832030892372131
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9540231227874756
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9567534327507019
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9759632349014282
},
{
"original": "VERGEZ Magali Signé",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.8517619967460632
},
{
"original": "Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.997373640537262
},
{
"original": "Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9975213408470154
},
{
"original": "VERGEZ Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9632388353347778
},
{
"original": "VERGEZ Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.927723228931427
},
{
"original": "VERGEZ Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9761468172073364
},
{
"original": "VERGEZ Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9799485802650452
},
{
"original": "VERGEZ Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9775839447975159
},
{
"original": "VERGEZ Magali Glyc",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9886924028396606
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.975821852684021
},
{
"original": "GLYCEMIE CAPILLAIRE",
"replacement": "[PERSONNE_2]",
"source": "ner",
"score": 0.9842458963394165
},
{
"original": "GLYCEMIE CAPILLAIRE",
"replacement": "[PERSONNE_2]",
"source": "ner",
"score": 0.9778326749801636
},
{
"original": "Jeanne Signé",
"replacement": "[PERSONNE_3]",
"source": "ner",
"score": 0.8135097622871399
},
{
"original": "Jeanne Signé",
"replacement": "[PERSONNE_3]",
"source": "ner",
"score": 0.9220442771911621
},
{
"original": "Jeanne",
"replacement": "[PERSONNE_3]",
"source": "ner",
"score": 0.8244627714157104
},
{
"original": "DECLERCK Marie",
"replacement": "[MEDECIN_6]",
"source": "ner",
"score": 0.8256687521934509
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9558121562004089
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.975185215473175
},
{
"original": "ENTRESTO",
"replacement": "[PERSONNE_4]",
"source": "ner",
"score": 0.8045708537101746
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9817212820053101
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9647781252861023
},
{
"original": "DIAZEPAM",
"replacement": "[PERSONNE_5]",
"source": "ner",
"score": 0.81693434715271
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.924862265586853
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.8739690184593201
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.8076850175857544
},
{
"original": "VERGEZ Magali",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.9772456884384155
},
{
"original": "VERGEZ Magali Signé",
"replacement": "[MEDECIN_3]",
"source": "ner",
"score": 0.8201007843017578
},
{
"original": "Léo SCHURDI-LEVRAUD",
"replacement": "[PERSONNE_6]",
"source": "ner",
"score": 0.9910992383956909
},
{
"original": "P.Larrouy",
"replacement": "[PERSONNE_7]",
"source": "ner",
"score": 0.9977959394454956
},
{
"original": "P.Larrouy",
"replacement": "[PERSONNE_7]",
"source": "ner",
"score": 0.9971214532852173
},
{
"original": "L Mascle",
"replacement": "[MEDECIN_1]",
"source": "ner",
"score": 0.9845480918884277
},
{
"original": "Fille priot",
"replacement": "[PATIENT_1]",
"source": "ner",
"score": 0.9773147106170654
},
{
"original": "Fille mme priot",
"replacement": "[PATIENT_1]",
"source": "ner",
"score": 0.967074990272522
}
]
}