#!/usr/bin/env python3
"""
Stroke Emergency Care Dataset
=================================
Each record = ONE acute stroke presentation.

Literature-Grounded Parameterization:
[1] Hertz JT et al. (2018). Stroke treatments, disability, mortality
    in urban Tanzania. Int J Qual Health Care.
    DOI: 10.1093/intqhc/mzy183
    KEY FINDINGS:
    - 149 subjects: 46% ischemic, 54% hemorrhagic
    - Mean age 57; 48% female; median NIHSS 19
    - No thrombolysis or thrombectomy available
    - Treatments: dysphagia screening 80%, aspirin 83%,
      antihypertensives 89%, statins 95%, DVT prophylaxis 0%
    - Ischemic: 19% died, 56% severe disability at discharge; 49% dead by 90d
    - Hemorrhagic: 33% died, 49% severe disability at discharge; 50% dead by 90d

[2] Bateman BT et al. (2013). Thrombolysis for acute ischemic stroke
    in South Africa. Int J Stroke. DOI: 10.1111/j.1747-4949.2012.00835.x
    - SICH 4.8%, overall mortality 7.1%
    - Safety/efficacy similar to developed countries

[3] Sarfo FS et al. (2017). Towards best practice acute stroke care
    Ghana. BMC Health Serv Res. DOI: 10.1186/s12913-017-2061-9
    - Limited/variable evidence-based stroke services
    - Low priority for stroke care in resource-poor settings

[4] Owolabi MO et al. (2022). Current management acute ischemic stroke
    Africa: review. Eur J Neurol. DOI: 10.1111/ene.15498
    - Lack of imaging, personnel, stroke units
    - Poor public awareness → delayed presentation
    - Telehealth suggested to mitigate scarcity

[5] Ogeng'o JA et al. (2014). Stroke in Africa. Afr J Neurol Sci.
    - Hemorrhagic stroke higher proportion (30-50%) vs HIC (15-20%)
    - Hypertension main risk factor (60-80%)
    - Younger mean age (55-60 vs 70+ in HIC)
"""
import numpy as np, pandas as pd, argparse, os

SCENARIOS = {
    'stroke_unit_available': {
        'exemplar': 'South Africa/Kenya (private)/Mauritius',
        'ct_scan_available': 0.80,    # [2] SA has CT
        'ct_within_1hr': 0.35,
        'thrombolysis_given': 0.08,   # [2] limited but available
        'thrombectomy': 0.02,
        'stroke_unit': 0.25,
        'neurologist_available': 0.30,
        'aspirin_given': 0.80,        # [1] 83%
        'antihypertensive': 0.85,     # [1] 89%
        'dvt_prophylaxis': 0.35,      # [1] 0% Tanzania; SA better
        'dysphagia_screening': 0.65,  # [1] 80%
        'rehabilitation': 0.30,
        'in_hospital_mortality': 0.18,
        'hemorrhagic_pct': 0.35,      # [5] higher than HIC
    },
    'basic_stroke_care': {
        'exemplar': 'Kenya (public)/Ghana/Tanzania/Ethiopia',
        'ct_scan_available': 0.40,
        'ct_within_1hr': 0.10,
        'thrombolysis_given': 0.01,   # [1] none at MNH
        'thrombectomy': 0.00,
        'stroke_unit': 0.05,
        'neurologist_available': 0.10,
        'aspirin_given': 0.65,
        'antihypertensive': 0.75,
        'dvt_prophylaxis': 0.05,      # [1] 0% Tanzania
        'dysphagia_screening': 0.40,
        'rehabilitation': 0.10,
        'in_hospital_mortality': 0.28, # [1] ischemic 19%, hemorrhagic 33%
        'hemorrhagic_pct': 0.45,      # [1] 54% at MNH
    },
    'no_stroke_services': {
        'exemplar': 'DRC/CAR/Sierra Leone/Niger',
        'ct_scan_available': 0.08,
        'ct_within_1hr': 0.02,
        'thrombolysis_given': 0.00,
        'thrombectomy': 0.00,
        'stroke_unit': 0.00,
        'neurologist_available': 0.02,
        'aspirin_given': 0.30,
        'antihypertensive': 0.40,
        'dvt_prophylaxis': 0.02,
        'dysphagia_screening': 0.10,
        'rehabilitation': 0.03,
        'in_hospital_mortality': 0.40,
        'hemorrhagic_pct': 0.50,
    },
}

def generate_dataset(n=10000, seed=42, scenario='basic_stroke_care'):
    rng = np.random.default_rng(seed)
    sc = SCENARIOS[scenario]
    records = []
    for idx in range(n):
        rec = {'id': idx + 1}
        rec['age'] = int(np.clip(rng.normal(58, 14), 20, 95))  # [1][5] younger
        rec['sex'] = rng.choice(['male','female'], p=[0.52, 0.48])  # [1] 48% female

        # Risk factors — [5] hypertension 60-80%
        rec['hypertension'] = 1 if rng.random() < 0.68 else 0
        rec['diabetes'] = 1 if rng.random() < 0.22 else 0
        rec['atrial_fibrillation'] = 1 if rng.random() < 0.08 else 0
        rec['smoking'] = 1 if rng.random() < 0.18 else 0
        rec['hiv_positive'] = 1 if rng.random() < 0.07 else 0
        rec['prior_stroke'] = 1 if rng.random() < 0.12 else 0
        rec['alcohol_heavy'] = 1 if rng.random() < 0.15 else 0
        rec['sickle_cell'] = 1 if rng.random() < 0.02 else 0

        # Stroke type — [1][5]
        rec['stroke_type'] = rng.choice(['ischemic','hemorrhagic'],
            p=[1-sc['hemorrhagic_pct'], sc['hemorrhagic_pct']])

        # Presentation
        rec['onset_to_arrival_hrs'] = round(max(0.5, rng.lognormal(
            np.log(12), 0.9)), 1)
        rec['arrived_within_4_5hrs'] = 1 if rec['onset_to_arrival_hrs'] <= 4.5 else 0
        rec['arrived_by_ems'] = 1 if rng.random() < 0.08 else 0
        rec['wake_up_stroke'] = 1 if rng.random() < 0.20 else 0
        rec['nihss_score'] = int(np.clip(rng.lognormal(np.log(12), 0.7), 1, 42))  # [1] median 19
        rec['gcs'] = int(np.clip(rng.choice([15,14,12,10,8,6,3],
            p=[0.25,0.15,0.15,0.15,0.12,0.10,0.08]), 3, 15))
        rec['systolic_bp'] = int(np.clip(rng.normal(170, 30), 90, 280))  # [5] hypertension
        rec['blood_glucose'] = round(np.clip(rng.lognormal(np.log(7), 0.4), 3, 30), 1)

        # Diagnostics — [1][4]
        rec['ct_scan_done'] = 1 if rng.random() < sc['ct_scan_available'] else 0
        rec['ct_within_1hr'] = 1 if (rec['ct_scan_done'] and rng.random() < sc['ct_within_1hr']) else 0
        rec['stroke_type_confirmed'] = 1 if rec['ct_scan_done'] else 0
        rec['ecg_done'] = 1 if rng.random() < sc['ct_scan_available'] * 0.8 else 0
        rec['echo_done'] = 1 if rng.random() < sc['ct_scan_available'] * 0.3 else 0
        rec['labs_complete'] = 1 if rng.random() < sc['ct_scan_available'] * 0.7 else 0

        # Treatment — [1][2]
        rec['thrombolysis_given'] = 1 if (
            rec['stroke_type'] == 'ischemic' and rec['ct_scan_done'] and
            rec['arrived_within_4_5hrs'] and rng.random() < sc['thrombolysis_given'] * 5
        ) else 0
        rec['thrombectomy_done'] = 1 if (
            rec['stroke_type'] == 'ischemic' and rng.random() < sc['thrombectomy']
        ) else 0
        rec['aspirin_given'] = 1 if (rec['stroke_type'] == 'ischemic' and rng.random() < sc['aspirin_given']) else 0
        rec['antihypertensive_given'] = 1 if rng.random() < sc['antihypertensive'] else 0
        rec['statin_given'] = 1 if rng.random() < sc['aspirin_given'] * 0.9 else 0
        rec['dvt_prophylaxis'] = 1 if rng.random() < sc['dvt_prophylaxis'] else 0
        rec['dysphagia_screening'] = 1 if rng.random() < sc['dysphagia_screening'] else 0
        rec['ngt_feeding'] = 1 if (not rec['dysphagia_screening'] and rec['gcs'] < 12 and rng.random() < 0.20) else 0
        rec['stroke_unit_admitted'] = 1 if rng.random() < sc['stroke_unit'] else 0
        rec['neurologist_consulted'] = 1 if rng.random() < sc['neurologist_available'] else 0
        rec['surgical_evacuation'] = 1 if (
            rec['stroke_type'] == 'hemorrhagic' and rec['nihss_score'] > 20 and rng.random() < sc['neurologist_available'] * 0.3
        ) else 0

        # Complications
        rec['aspiration_pneumonia'] = 1 if (not rec['dysphagia_screening'] and rng.random() < 0.15) else 0
        rec['dvt_pe'] = 1 if (not rec['dvt_prophylaxis'] and rng.random() < 0.08) else 0
        rec['hemorrhagic_transformation'] = 1 if (rec['stroke_type'] == 'ischemic' and rec['thrombolysis_given'] and rng.random() < 0.05) else 0
        rec['seizure'] = 1 if rng.random() < 0.08 else 0
        rec['raised_icp'] = 1 if (rec['stroke_type'] == 'hemorrhagic' and rec['nihss_score'] > 15 and rng.random() < 0.25) else 0

        # Outcome — [1]
        mort = sc['in_hospital_mortality']
        mort *= 1.5 if rec['stroke_type'] == 'hemorrhagic' else 0.7  # [1]
        mort *= (rec['nihss_score'] / 15)  # severity
        mort *= 0.5 if rec['stroke_unit_admitted'] else 1
        mort *= 0.6 if rec['thrombolysis_given'] else 1
        mort *= 1.5 if rec['aspiration_pneumonia'] else 1
        mort *= 1.3 if rec['gcs'] < 9 else 1
        rec['died_in_hospital'] = 1 if rng.random() < min(0.85, mort) else 0
        rec['mrs_discharge'] = int(rng.choice([0,1,2,3,4,5,6],
            p=[0.02,0.05,0.08,0.15,0.25,0.25,0.20])) if not rec['died_in_hospital'] else 6
        rec['rehabilitation_received'] = 1 if (not rec['died_in_hospital'] and rng.random() < sc['rehabilitation']) else 0
        rec['los_days'] = max(1, int(rng.lognormal(np.log(7), 0.6))) if not rec['died_in_hospital'] else 0

        rec['year'] = rng.choice([2019,2020,2021,2022,2023], p=[0.12,0.18,0.20,0.25,0.25])
        records.append(rec)

    df = pd.DataFrame(records)
    print(f"\n{'='*60}\nStroke Emergency Care — {scenario} ({sc['exemplar']})")
    print(f"  Hemorrhagic: {(df['stroke_type']=='hemorrhagic').mean()*100:.1f}% | CT done: {df['ct_scan_done'].mean()*100:.1f}%")
    print(f"  Thrombolysis: {df['thrombolysis_given'].mean()*100:.1f}% | Stroke unit: {df['stroke_unit_admitted'].mean()*100:.1f}%")
    print(f"  In-hospital mortality: {df['died_in_hospital'].mean()*100:.1f}%")
    return df

if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('--all-scenarios', action='store_true')
    parser.add_argument('--n', type=int, default=10000)
    parser.add_argument('--seed', type=int, default=42)
    args = parser.parse_args()
    os.makedirs('data', exist_ok=True)
    if args.all_scenarios:
        for sc in SCENARIOS:
            df = generate_dataset(n=args.n, seed=args.seed, scenario=sc)
            df.to_csv(os.path.join('data', f'stroke_{sc}.csv'), index=False)
            print(f"  -> Saved\n")
    else:
        df = generate_dataset(n=args.n, seed=args.seed)
        df.to_csv(os.path.join('data', 'stroke_basic_stroke_care.csv'), index=False)