PTSRaiAgent/app/llm/agent/drug_safety.py

"""Deterministic drug-safety lookup: composition, contraindications, interactions, patient-specific findings.

The LLM never decides whether a drug is safe for the patient (ADR-007 spirit): it calls `check_patient` and
retells the findings. Matching is by tags shared with the ЕМИАС record (`data/emias/records.yaml`):
condition tags → `cautions[].when_condition`, allergy tags → `composition.excipient_tags`, other medication
classes → `interactions[].with_class`. Source data: `data/meds/drug_safety.yaml` (prototype retelling of ГРЛС).
"""

from functools import lru_cache
from pathlib import Path
from typing import Literal

import yaml
from pydantic import BaseModel, Field

from app.config import get_settings
from app.integrations.emias_record import PatientRecord

Severity = Literal["high", "medium", "low"]


class Caution(BaseModel):
    when_condition: str
    severity: Severity
    text: str


class Interaction(BaseModel):
    with_class: str
    severity: Severity
    text: str


class Composition(BaseModel):
    form: str
    active: str
    excipients: list[str] = Field(default_factory=list)
    excipient_tags: list[str] = Field(default_factory=list)
    note: str = ""


class DrugEntry(BaseModel):
    inn: str
    aliases: list[str] = Field(default_factory=list)
    drug_class: str = ""
    classes: list[str] = Field(default_factory=list)  # classes this drug belongs to (e.g. nsaid, ssri)
    contraindications: list[str] = Field(default_factory=list)
    cautions: list[Caution] = Field(default_factory=list)
    interactions: list[Interaction] = Field(default_factory=list)
    composition: Composition | None = None
    pregnancy: str = ""
    source: str = ""


class Finding(BaseModel):
    severity: Severity
    kind: Literal["condition", "allergy", "interaction", "pregnancy", "contraindication"]
    text: str


class SafetyDb(BaseModel):
    drugs: dict[str, DrugEntry]
    allergy_labels: dict[str, str] = Field(default_factory=dict)


def _path() -> Path:
    return get_settings().data_dir / "meds" / "drug_safety.yaml"


@lru_cache
def load_db(path: Path | None = None) -> SafetyDb:
    raw = yaml.safe_load((path or _path()).read_text(encoding="utf-8"))
    classes_of: dict[str, list[str]] = {}
    for cls, members in (raw.get("drug_classes") or {}).items():
        for m in members:
            classes_of.setdefault(m, []).append(cls)
    drugs: dict[str, DrugEntry] = {}
    for inn, d in (raw.get("drugs") or {}).items():
        entry = DrugEntry(
            inn=inn,
            aliases=d.get("aliases", []),
            drug_class=d.get("class", ""),
            classes=sorted(set(d.get("classes", [])) | set(classes_of.get(inn, []))),
            contraindications=d.get("contraindications", []),
            cautions=d.get("cautions", []),
            interactions=d.get("interactions", []),
            composition=d.get("composition"),
            pregnancy=d.get("pregnancy", ""),
            source=d.get("source", ""),
        )
        drugs[inn] = entry
    return SafetyDb(drugs=drugs, allergy_labels=raw.get("allergy_labels") or {})


def find_drug(name: str, db: SafetyDb | None = None) -> DrugEntry | None:
    """By INN or alias, case-insensitive; "сертралин 50 мг" → сертралин."""
    db = db or load_db()
    words = [w.strip(".,!?;:«»()") for w in name.lower().replace("ё", "е").split()]
    for entry in db.drugs.values():
        names = {entry.inn, *(a.lower() for a in entry.aliases)}
        # Russian case endings: "сертралине", "ибупрофена" — match on the stem (name minus the last letter).
        if any(w in names or any(len(n) > 4 and w.startswith(n[:-1]) for n in names) for w in words):
            return entry
    return None


def check_patient(drug: DrugEntry, record: PatientRecord, db: SafetyDb | None = None) -> list[Finding]:
    """Patient-specific findings for `drug`, most severe first. Empty list = nothing found in the reference —
    which is NOT a statement that the drug is safe (the answer must say so)."""
    db = db or load_db()
    out: list[Finding] = []
    cond_tags = {t for c in record.conditions for t in c.tags}
    for c in drug.cautions:
        if c.when_condition in cond_tags:
            out.append(Finding(severity=c.severity, kind="condition", text=c.text))
    # Excipients vs allergies / intolerances.
    allergy_tags = {t for a in record.allergies for t in a.tags}
    if drug.composition is not None:
        for tag in sorted(allergy_tags & set(drug.composition.excipient_tags)):
            label = db.allergy_labels.get(tag, tag)
            out.append(
                Finding(
                    severity="high", kind="allergy", text=f"В составе есть {label}, а в карте отмечена непереносимость."
                )
            )
    # Interactions with the patient's other medicines and with the current psychiatric prescription.
    other_classes: dict[str, str] = {}
    for m in record.other_medications:
        for cls in m.classes:
            other_classes[cls] = m.drug
    for rx in record.current_prescriptions:
        entry = find_drug(rx.drug, db)
        if entry is not None and entry.inn != drug.inn:
            for cls in entry.classes:
                other_classes.setdefault(cls, rx.drug)
    for i in drug.interactions:
        if i.with_class in other_classes:
            out.append(
                Finding(
                    severity=i.severity, kind="interaction", text=f"{i.text} (в карте: {other_classes[i.with_class]})."
                )
            )
    if record.pregnancy and drug.pregnancy:
        out.append(Finding(severity="medium", kind="pregnancy", text=drug.pregnancy))
    order = {"high": 0, "medium": 1, "low": 2}
    out.sort(key=lambda f: order[f.severity])
    return out