Add 15 drug profiles to pharma KB from Kaggle dataset

knowledge/pharma/features.yaml

@@ -96,3 +96,93 @@ documents:
       standard treatment at scale. Enrollment eligibility is defined by inclusion
       and exclusion criteria in the protocol.
     tags: [clinical-trial, study, investigational]
+
+  - id: pharma_drug_001
+    title: "Seroxetine50 — Drug Profile"
+    content: >
+      Seroxetine50 is indicated for the treatment or management of Allergy relief. Known adverse events include: Fatigue, Nausea. Contraindicated in patients with: Bleeding disorders. Prescriber warning: Avoid alcohol. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["seroxetine50", "allergy-relief", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_002
+    title: "Mecoparin93 — Drug Profile"
+    content: >
+      Mecoparin93 is indicated for the treatment or management of Allergy relief. Known adverse events include: Nausea. Contraindicated in patients with: Allergic reaction. Prescriber warning: Take with food. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["mecoparin93", "allergy-relief", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_003
+    title: "Daxozole89 — Drug Profile"
+    content: >
+      Daxozole89 is indicated for the treatment or management of Allergy relief. Known adverse events include: Diarrhea, Blurred vision, Dizziness. Contraindicated in patients with: High blood pressure. Prescriber warning: Take with food. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["daxozole89", "allergy-relief", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_004
+    title: "Viracillin84 — Drug Profile"
+    content: >
+      Viracillin84 is indicated for the treatment or management of Inflammation reduction. Known adverse events include: Fatigue, Dry mouth. Contraindicated in patients with: Kidney impairment. Prescriber warning: Do not operate machinery. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["viracillin84", "inflammation-reduction", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_005
+    title: "Amoxstatin62 — Drug Profile"
+    content: >
+      Amoxstatin62 is indicated for the treatment or management of Psychosis control. Known adverse events include: Insomnia, Dry mouth, Fatigue. Contraindicated in patients with: Bleeding disorders. Prescriber warning: Do not operate machinery. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["amoxstatin62", "psychosis-control", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_006
+    title: "Loxocillin72 — Drug Profile"
+    content: >
+      Loxocillin72 is indicated for the treatment or management of Viral infections. Known adverse events include: Rash, Dizziness. Contraindicated in patients with: Bleeding disorders. Prescriber warning: May affect fertility. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["loxocillin72", "viral-infections", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_007
+    title: "Fenoxetine43 — Drug Profile"
+    content: >
+      Fenoxetine43 is indicated for the treatment or management of Psychosis control. Known adverse events include: Blurred vision. Contraindicated in patients with: Bleeding disorders. Prescriber warning: May affect fertility. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["fenoxetine43", "psychosis-control", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_008
+    title: "Fenomab17 — Drug Profile"
+    content: >
+      Fenomab17 is indicated for the treatment or management of Bacterial infections. Known adverse events include: Nausea. Contraindicated in patients with: Seizures. Prescriber warning: Avoid alcohol. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["fenomab17", "bacterial-infections", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_009
+    title: "Cefparin42 — Drug Profile"
+    content: >
+      Cefparin42 is indicated for the treatment or management of Prevent infectious disease. Known adverse events include: Dry mouth. Contraindicated in patients with: Asthma. Prescriber warning: Avoid alcohol. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["cefparin42", "prevent-infectious-disease", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_010
+    title: "Rivaprofen81 — Drug Profile"
+    content: >
+      Rivaprofen81 is indicated for the treatment or management of Bacterial infections. Known adverse events include: Insomnia, Fatigue, Headache. Contraindicated in patients with: Pregnancy. Prescriber warning: Take with food. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["rivaprofen81", "bacterial-infections", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_011
+    title: "Fenocillin43 — Drug Profile"
+    content: >
+      Fenocillin43 is indicated for the treatment or management of Allergy relief. Known adverse events include: Dizziness, Fatigue. Contraindicated in patients with: Liver disease. Prescriber warning: Risk of addiction. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["fenocillin43", "allergy-relief", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_012
+    title: "Amoxxetine60 — Drug Profile"
+    content: >
+      Amoxxetine60 is indicated for the treatment or management of Allergy relief. Known adverse events include: Dry mouth. Contraindicated in patients with: Allergic reaction. Prescriber warning: Do not operate machinery. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["amoxxetine60", "allergy-relief", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_013
+    title: "Zylomab54 — Drug Profile"
+    content: >
+      Zylomab54 is indicated for the treatment or management of Fungal infections. Known adverse events include: Rash, Diarrhea, Insomnia. Contraindicated in patients with: Liver disease. Prescriber warning: Monitor blood pressure. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["zylomab54", "fungal-infections", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_014
+    title: "Amoxstatin52 — Drug Profile"
+    content: >
+      Amoxstatin52 is indicated for the treatment or management of Depression treatment. Known adverse events include: Constipation, Diarrhea, Blurred vision. Contraindicated in patients with: Pregnancy. Prescriber warning: Do not operate machinery. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["amoxstatin52", "depression-treatment", "adverse-event", "drug-profile"]
+
+  - id: pharma_drug_015
+    title: "Virazole50 — Drug Profile"
+    content: >
+      Virazole50 is indicated for the treatment or management of Viral infections. Known adverse events include: Headache, Dizziness, Constipation. Contraindicated in patients with: Seizures. Prescriber warning: May cause drowsiness. Serious unexpected adverse events must be reported to the regulatory authority within 15 days.
+    tags: ["virazole50", "viral-infections", "adverse-event", "drug-profile"]

scripts/kaggle_to_yaml.py

@@ -24,9 +24,12 @@ from pathlib import Path
 KNOWLEDGE_ROOT = Path(__file__).parent.parent / "knowledge"
 
 _DRUG_COLS    = ["drug_name", "Drug Name", "name", "drug", "drugName"]
-_USE_COLS     = ["medical_condition", "condition", "use", "uses", "indication", "Medical Condition"]
+_USE_COLS     = ["indications", "medical_condition", "condition", "use", "uses",
+                 "indication", "Medical Condition"]
 _EFFECT_COLS  = ["side_effects", "Side_Effects", "Side Effects", "sideEffects",
                  "adverse_effects", "adverse_events"]
+_CONTRA_COLS  = ["contraindications", "contraindication", "Contraindications"]
+_WARN_COLS    = ["warnings", "warning", "Warnings", "precautions"]
 
 
 def _find_col(headers: list[str], candidates: list[str]) -> str | None:
@@ -71,12 +74,15 @@ def convert(csv_path: Path, max_drugs: int, dry_run: bool, domain: str) -> None:
         drug_col   = _find_col(headers, _DRUG_COLS)
         use_col    = _find_col(headers, _USE_COLS)
         effect_col = _find_col(headers, _EFFECT_COLS)
+        contra_col = _find_col(headers, _CONTRA_COLS)
+        warn_col   = _find_col(headers, _WARN_COLS)
 
         if not drug_col:
             print(f"ERROR: no drug name column found. Headers: {headers}", file=sys.stderr)
             sys.exit(1)
 
-        print(f"Detected columns — drug: {drug_col!r}, use: {use_col!r}, effects: {effect_col!r}")
+        print(f"Detected columns — drug: {drug_col!r}, use: {use_col!r}, "
+              f"effects: {effect_col!r}, contra: {contra_col!r}, warnings: {warn_col!r}")
 
         blocks: list[str] = []
         seen: set[str] = set()
@@ -92,15 +98,25 @@ def convert(csv_path: Path, max_drugs: int, dry_run: bool, domain: str) -> None:
 
             condition = row.get(use_col, "").strip() if use_col else ""
             effects   = row.get(effect_col, "").strip() if effect_col else ""
+            contra    = row.get(contra_col, "").strip() if contra_col else ""
+            warnings  = row.get(warn_col, "").strip() if warn_col else ""
+
             condition_str = condition or "the indicated condition"
             effects_str   = _truncate(effects) if effects else "not listed"
 
-            content = (
-                f"{drug} is indicated for the treatment or management of {condition_str}. "
-                f"Known adverse events associated with {drug} include: {effects_str}. "
-                f"Prescribers should monitor patients and report serious unexpected occurrences "
-                f"to the regulatory authority within 15 days."
+            parts = [
+                f"{drug} is indicated for the treatment or management of {condition_str}.",
+                f"Known adverse events include: {effects_str}.",
+            ]
+            if contra:
+                parts.append(f"Contraindicated in patients with: {contra}.")
+            if warnings:
+                parts.append(f"Prescriber warning: {warnings}.")
+            parts.append(
+                "Serious unexpected adverse events must be reported to the regulatory "
+                "authority within 15 days."
             )
+            content = " ".join(parts)
 
             tags = list(filter(None, [
                 _slugify(drug),