| """Generate synthetic chemical poisoning & toxicology dataset for SSA. |
| |
| Research-based parameterization: |
| - BMC Public Health (2023): Child/adolescent pesticide poisoning mortality |
| in SA; street pesticides (aldicarb, organophosphates) sold illegally. |
| - Frontiers (2023): Children poisoning in LMICs - 4x higher mortality; |
| medications, pesticides, kerosene, household chemicals. |
| - StatPearls: Organophosphate toxicity highest in agricultural developing |
| nations with less stringent regulations. |
| - Beyond Pesticides (2024): Dozens of children died in SA from |
| unregulated pesticide use in communities. |
| - WHO: Poisoning causes ~45,000 deaths/yr in Africa. |
| """ |
|
|
| from __future__ import annotations |
|
|
| from pathlib import Path |
|
|
| import numpy as np |
| import pandas as pd |
|
|
| SEED = 42 |
| N_PER_SCENARIO = 10_000 |
|
|
| YEAR_RANGE = np.arange(2010, 2025) |
| YEAR_WEIGHTS = np.linspace(0.85, 1.3, len(YEAR_RANGE)) |
| YEAR_WEIGHTS = YEAR_WEIGHTS / YEAR_WEIGHTS.sum() |
|
|
| SCENARIOS = { |
| "agricultural_pesticide": { |
| "setting_probs": {"rural_farm": 0.50, "peri_urban": 0.30, "urban": 0.20}, |
| "agent_probs": {"organophosphate": 0.35, "carbamate": 0.20, "pyrethroid": 0.15, |
| "herbicide": 0.15, "rodenticide": 0.10, "fungicide": 0.05}, |
| "intent_probs": {"accidental_occupational": 0.40, "accidental_child": 0.20, |
| "intentional_self_harm": 0.25, "intentional_other": 0.05, "unknown": 0.10}, |
| "mortality_rate": 0.08, |
| "child_pct": 0.25, |
| "ppe_use_pct": 0.12, |
| "poison_centre_access": 0.05, |
| "antidote_available": 0.30, |
| }, |
| "household_chemical_urban": { |
| "setting_probs": {"urban_informal": 0.40, "urban_formal": 0.30, "peri_urban": 0.30}, |
| "agent_probs": {"kerosene_paraffin": 0.30, "bleach_caustic": 0.20, |
| "medication_overdose": 0.20, "rat_poison_street": 0.15, |
| "traditional_medicine": 0.10, "other_chemical": 0.05}, |
| "intent_probs": {"accidental_child": 0.40, "accidental_adult": 0.15, |
| "intentional_self_harm": 0.30, "intentional_other": 0.05, "unknown": 0.10}, |
| "mortality_rate": 0.05, |
| "child_pct": 0.45, |
| "ppe_use_pct": 0.0, |
| "poison_centre_access": 0.10, |
| "antidote_available": 0.40, |
| }, |
| "industrial_occupational": { |
| "setting_probs": {"industrial": 0.45, "mining": 0.25, "construction": 0.15, "urban": 0.15}, |
| "agent_probs": {"solvent_hydrocarbon": 0.25, "heavy_metal_compound": 0.20, |
| "acid_alkali": 0.15, "gas_fume": 0.20, |
| "pesticide_industrial": 0.10, "other_industrial": 0.10}, |
| "intent_probs": {"accidental_occupational": 0.65, "accidental_other": 0.15, |
| "intentional_self_harm": 0.10, "intentional_other": 0.02, "unknown": 0.08}, |
| "mortality_rate": 0.06, |
| "child_pct": 0.05, |
| "ppe_use_pct": 0.20, |
| "poison_centre_access": 0.08, |
| "antidote_available": 0.35, |
| }, |
| } |
|
|
| SCENARIO_FILES = { |
| "agricultural_pesticide": "poisoning_agricultural.csv", |
| "household_chemical_urban": "poisoning_household.csv", |
| "industrial_occupational": "poisoning_industrial.csv", |
| } |
|
|
| ROUTES = {"ingestion": 0.50, "dermal": 0.20, "inhalation": 0.20, "injection": 0.02, "ocular": 0.08} |
| SEVERITY = {"mild": 0.35, "moderate": 0.35, "severe": 0.20, "fatal": 0.10} |
|
|
|
|
| def _choice(rng, prob_map): |
| keys = list(prob_map.keys()) |
| weights = np.array(list(prob_map.values()), dtype=float) |
| weights = weights / weights.sum() |
| return rng.choice(keys, p=weights) |
|
|
|
|
| def _simulate_scenario(name, params, seed): |
| rng = np.random.default_rng(seed) |
| records = [] |
|
|
| for idx in range(N_PER_SCENARIO): |
| year = int(rng.choice(YEAR_RANGE, p=YEAR_WEIGHTS)) |
| setting = _choice(rng, params["setting_probs"]) |
| is_child = int(rng.random() < params["child_pct"]) |
| age = int(np.clip(rng.normal(3, 1.5) if is_child else rng.normal(32, 12), 0, 70)) |
| sex = rng.choice(["male", "female"], p=[0.55, 0.45]) |
|
|
| agent = _choice(rng, params["agent_probs"]) |
| intent = _choice(rng, params["intent_probs"]) |
| route = _choice(rng, ROUTES) |
|
|
| time_to_presentation_hrs = float(np.clip(rng.lognormal(np.log(3), 0.8), 0.5, 72)) |
| delayed_presentation = int(time_to_presentation_hrs > 6) |
|
|
| severity = _choice(rng, SEVERITY) |
| if intent == "intentional_self_harm": |
| severity = rng.choice(["moderate", "severe", "fatal"], p=[0.30, 0.45, 0.25]) |
|
|
| |
| gi_symptoms = int(route == "ingestion" and rng.random() < 0.70) |
| respiratory_distress = int(route == "inhalation" and rng.random() < 0.50) |
| cholinergic_crisis = int(agent in ("organophosphate", "carbamate") and rng.random() < 0.45) |
| seizures = int(severity in ("severe", "fatal") and rng.random() < 0.15) |
| altered_consciousness = int(severity in ("severe", "fatal") and rng.random() < 0.30) |
| chemical_burn = int(agent in ("acid_alkali", "bleach_caustic") and rng.random() < 0.40) |
| aspiration_pneumonia = int(agent == "kerosene_paraffin" and rng.random() < 0.25) |
|
|
| |
| ppe_use = int(rng.random() < params["ppe_use_pct"]) |
| decontamination = int(rng.random() < 0.40) |
| activated_charcoal = int(route == "ingestion" and time_to_presentation_hrs < 2 and rng.random() < 0.30) |
| antidote_available = int(rng.random() < params["antidote_available"]) |
| antidote_given = int(antidote_available and severity in ("moderate", "severe", "fatal") and rng.random() < 0.70) |
| atropine_given = int(cholinergic_crisis and rng.random() < 0.60) |
| icu_admission = int(severity in ("severe", "fatal") and rng.random() < 0.40) |
| ventilator = int(icu_admission and rng.random() < 0.30) |
|
|
| poison_centre_consulted = int(rng.random() < params["poison_centre_access"]) |
| referred_higher = int(severity in ("severe", "fatal") and rng.random() < 0.50) |
|
|
| |
| died = int(severity == "fatal" and rng.random() < params["mortality_rate"] * 10) |
| sequelae = int(severity in ("severe", "fatal") and not died and rng.random() < 0.15) |
| hospital_days = int(np.clip( |
| rng.poisson(1 if severity == "mild" else 3 if severity == "moderate" else 7), 0, 30)) |
|
|
| |
| safe_storage = int(rng.random() < 0.20) |
| child_proof_container = int(is_child and rng.random() < 0.05) |
| labelled_container = int(rng.random() < 0.30) |
| pesticide_regulation = int(rng.random() < 0.15) |
|
|
| record = { |
| "record_id": f"{name[:3].upper()}-{idx:05d}", |
| "scenario": name, |
| "year": year, |
| "setting": setting, |
| "age": age, |
| "sex": sex, |
| "is_child": is_child, |
| "agent": agent, |
| "intent": intent, |
| "route": route, |
| "time_to_presentation_hrs": round(time_to_presentation_hrs, 1), |
| "delayed_presentation": delayed_presentation, |
| "severity": severity, |
| "gi_symptoms": gi_symptoms, |
| "respiratory_distress": respiratory_distress, |
| "cholinergic_crisis": cholinergic_crisis, |
| "seizures": seizures, |
| "altered_consciousness": altered_consciousness, |
| "chemical_burn": chemical_burn, |
| "aspiration_pneumonia": aspiration_pneumonia, |
| "ppe_use": ppe_use, |
| "decontamination": decontamination, |
| "activated_charcoal": activated_charcoal, |
| "antidote_available": antidote_available, |
| "antidote_given": antidote_given, |
| "atropine_given": atropine_given, |
| "icu_admission": icu_admission, |
| "ventilator": ventilator, |
| "poison_centre_consulted": poison_centre_consulted, |
| "referred_higher": referred_higher, |
| "died": died, |
| "sequelae": sequelae, |
| "hospital_days": hospital_days, |
| "safe_storage": safe_storage, |
| "child_proof_container": child_proof_container, |
| "labelled_container": labelled_container, |
| "pesticide_regulation": pesticide_regulation, |
| } |
| records.append(record) |
|
|
| return pd.DataFrame(records) |
|
|
|
|
| def main(): |
| output_dir = Path("data") |
| output_dir.mkdir(parents=True, exist_ok=True) |
| for idx, (name, params) in enumerate(SCENARIOS.items()): |
| df = _simulate_scenario(name, params, SEED + idx * 211) |
| df.to_csv(output_dir / SCENARIO_FILES[name], index=False) |
| print(f"Saved {name} -> {SCENARIO_FILES[name]}") |
|
|
|
|
| if __name__ == "__main__": |
| main() |
|
|
