Spaces:
Running
Running
File size: 9,252 Bytes
877add7 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 | #!/usr/bin/env python3
"""Build required processed data artifacts for POLYGUARD-OPENENV."""
from __future__ import annotations
import json
from datetime import datetime, timezone
from pathlib import Path
import sys
ROOT = Path(__file__).resolve().parents[1]
if str(ROOT) not in sys.path:
sys.path.insert(0, str(ROOT))
from typing import Any
import pandas as pd
import yaml
from app.knowledge.ddi_knowledge import is_contraindicated_pair
from app.knowledge.drug_catalog import DRUG_CLASSES
from app.knowledge.substitution_rules import SUBSTITUTIONS
from app.knowledge.taper_rules import requires_taper
def _safe_write_json(path: Path, payload: Any) -> None:
path.parent.mkdir(parents=True, exist_ok=True)
path.write_text(json.dumps(payload, ensure_ascii=True, indent=2), encoding="utf-8")
def _write_jsonl(path: Path, rows: list[dict[str, Any]]) -> None:
path.parent.mkdir(parents=True, exist_ok=True)
with path.open("w", encoding="utf-8") as f:
for row in rows:
f.write(json.dumps(row, ensure_ascii=True) + "\n")
def _load_scenario_rows(scenario_dir: Path) -> list[dict[str, Any]]:
rows: list[dict[str, Any]] = []
if not scenario_dir.exists():
return rows
for path in sorted(scenario_dir.glob("*.json")):
rows.append(json.loads(path.read_text(encoding="utf-8")))
return rows
def main() -> None:
root = Path(__file__).resolve().parents[1]
processed_dir = root / "data" / "processed"
processed_dir.mkdir(parents=True, exist_ok=True)
artifacts_dir = root / "data" / "artifacts"
artifacts_dir.mkdir(parents=True, exist_ok=True)
drug_rows: list[dict[str, Any]] = []
class_rows: list[dict[str, Any]] = []
for idx, (drug, class_name) in enumerate(sorted(DRUG_CLASSES.items()), start=1):
canonical_id = f"drug_{idx:04d}"
aliases = [drug.replace("_", " "), drug.upper()]
drug_rows.append(
{
"canonical_id": canonical_id,
"canonical_name": drug,
"aliases": aliases,
"class_name": class_name,
"source": "local_drug_catalog",
}
)
class_rows.append(
{
"canonical_id": canonical_id,
"class_name": class_name,
"subclass": f"{class_name}_core",
"source": "local_drug_catalog",
}
)
interactions: list[dict[str, Any]] = []
drugs = sorted(DRUG_CLASSES)
for i, drug_a in enumerate(drugs):
for drug_b in drugs[i + 1 :]:
if is_contraindicated_pair(drug_a, drug_b):
interactions.append(
{
"drug_a": drug_a,
"drug_b": drug_b,
"severity": "high",
"interaction_type": "contraindicated",
"source": "ddi_rules",
}
)
burden_rules = {
"version": "1.0",
"formula": "burden = med_count/12 + high_risk_count*0.04",
"high_risk_classes": ["sedative", "anticoagulant", "analgesic"],
}
taper_rows = [
{"drug": drug, "requires_taper": requires_taper(drug), "default_taper_days": 14 if requires_taper(drug) else 0}
for drug in drugs
]
taper_rules = {"rules": taper_rows, "source": "taper_rules"}
substitution_rules = {"rules": SUBSTITUTIONS, "source": "substitution_rules"}
retrieval_index_file = root / "data" / "retrieval_index" / "index.json"
retrieval_rows: list[dict[str, Any]] = []
if retrieval_index_file.exists():
retrieval_rows = json.loads(retrieval_index_file.read_text(encoding="utf-8"))
retrieval_corpus = [
{
"doc_id": row.get("id"),
"path": row.get("path"),
"text": row.get("text"),
"source": "retrieval_index",
}
for row in retrieval_rows
]
graph_edges: list[dict[str, Any]] = []
for drug, class_name in sorted(DRUG_CLASSES.items()):
graph_edges.append({"src": drug, "dst": class_name, "edge_type": "in_class", "weight": 1.0})
for row in interactions:
graph_edges.append({"src": row["drug_a"], "dst": row["drug_b"], "edge_type": "contraindicated_with", "weight": 1.0})
graph_edges.append({"src": row["drug_b"], "dst": row["drug_a"], "edge_type": "contraindicated_with", "weight": 1.0})
for src, replacements in SUBSTITUTIONS.items():
for dst in replacements:
graph_edges.append({"src": src, "dst": dst, "edge_type": "substitute_for", "weight": 0.8})
synthetic_file = root / "data" / "synthetic" / "synthetic_patients.json"
synthetic_rows: list[dict[str, Any]] = []
if synthetic_file.exists():
synthetic_rows = json.loads(synthetic_file.read_text(encoding="utf-8"))
easy_rows = _load_scenario_rows(root / "data" / "scenarios" / "easy")
medium_rows = _load_scenario_rows(root / "data" / "scenarios" / "medium")
hard_rows = _load_scenario_rows(root / "data" / "scenarios" / "hard")
pd.DataFrame(drug_rows).to_parquet(processed_dir / "normalized_drugs.parquet", index=False)
pd.DataFrame(class_rows).to_parquet(processed_dir / "drug_classes.parquet", index=False)
pd.DataFrame(interactions).to_parquet(processed_dir / "interactions.parquet", index=False)
pd.DataFrame(graph_edges).to_parquet(processed_dir / "graph_edges.parquet", index=False)
pd.DataFrame(synthetic_rows).to_parquet(processed_dir / "patients_synthetic.parquet", index=False)
(processed_dir / "burden_rules.yaml").write_text(yaml.safe_dump(burden_rules, sort_keys=False), encoding="utf-8")
(processed_dir / "taper_rules.yaml").write_text(yaml.safe_dump(taper_rules, sort_keys=False), encoding="utf-8")
(processed_dir / "substitution_rules.yaml").write_text(yaml.safe_dump(substitution_rules, sort_keys=False), encoding="utf-8")
_write_jsonl(processed_dir / "retrieval_corpus.jsonl", retrieval_corpus)
_write_jsonl(root / "data" / "scenarios" / "scenarios_easy.jsonl", easy_rows)
_write_jsonl(root / "data" / "scenarios" / "scenarios_medium.jsonl", medium_rows)
_write_jsonl(root / "data" / "scenarios" / "scenarios_hard.jsonl", hard_rows)
feature_dictionary = {
"normalized_drugs": ["canonical_id", "canonical_name", "aliases", "class_name", "source"],
"drug_classes": ["canonical_id", "class_name", "subclass", "source"],
"interactions": ["drug_a", "drug_b", "severity", "interaction_type", "source"],
"graph_edges": ["src", "dst", "edge_type", "weight"],
"patients_synthetic": [
"patient_id",
"age",
"sex",
"comorbidities",
"medications",
"labs",
"vitals",
"specialist_conflicts",
"prior_ade_history",
"frailty_score",
"adherence_estimate",
],
}
_safe_write_json(processed_dir / "feature_dictionary.json", feature_dictionary)
provenance_manifest = {
"generated_at": datetime.now(timezone.utc).isoformat(),
"policy": {
"core_sources_live_required": ["canonical_vocab", "interactions"],
"secondary_sources_fallback": True,
"weak_signal_labels_marked": True,
},
"inputs": {
"drug_catalog": "app/knowledge/drug_catalog.py",
"ddi_rules": "app/knowledge/ddi_knowledge.py",
"substitutions": "app/knowledge/substitution_rules.py",
"taper_rules": "app/knowledge/taper_rules.py",
"retrieval_index": str(retrieval_index_file),
},
"counts": {
"normalized_drugs": len(drug_rows),
"interactions": len(interactions),
"retrieval_docs": len(retrieval_corpus),
"scenario_easy": len(easy_rows),
"scenario_medium": len(medium_rows),
"scenario_hard": len(hard_rows),
"patients_synthetic": len(synthetic_rows),
},
}
_safe_write_json(processed_dir / "provenance_manifest.json", provenance_manifest)
dataset_report = f"""# Dataset Report
## Summary
- Normalized drugs: {len(drug_rows)}
- Drug classes: {len(class_rows)}
- Interactions: {len(interactions)}
- Graph edges: {len(graph_edges)}
- Synthetic patients: {len(synthetic_rows)}
- Scenarios (easy/medium/hard): {len(easy_rows)}/{len(medium_rows)}/{len(hard_rows)}
- Retrieval corpus documents: {len(retrieval_corpus)}
## Source Policy
- Core vocabulary/interactions are treated as core sources.
- Secondary sources are allowed fallback with explicit provenance.
- Weak/noisy safety signals are labeled as such in provenance metadata.
## Artifacts
Artifacts are stored under `data/processed`, `data/scenarios`, and `data/artifacts`.
"""
(root / "docs" / "dataset_report.md").write_text(dataset_report, encoding="utf-8")
summary = {
"status": "ok",
"processed_dir": str(processed_dir),
"docs_report": str(root / "docs" / "dataset_report.md"),
}
_safe_write_json(artifacts_dir / "bootstrap_data_summary.json", summary)
print("bootstrap_data_done")
if __name__ == "__main__":
main()
|