Add dataset files

README.md

@@ -0,0 +1,75 @@
+---
+license: cc-by-4.0
+task_categories:
+  - tabular-classification
+  - tabular-regression
+language:
+  - en
+tags:
+  - healthcare
+  - medicine-quality
+  - QC-laboratory
+  - ISO-17025
+  - WHO-prequalification
+  - HPLC
+  - proficiency-testing
+  - sub-saharan-africa
+  - lmic
+pretty_name: "Medicine Quality Control Laboratories (Equipment, Staff, Proficiency, Budget)"
+size_categories:
+  - 10K<n<100K
+configs:
+  - config_name: who_prequalified_lab
+    data_files: data/qc_lab_who_prequalified_lab.csv
+  - config_name: national_reference_lab
+    data_files: data/qc_lab_national_reference_lab.csv
+    default: true
+  - config_name: district_basic_lab
+    data_files: data/qc_lab_district_basic_lab.csv
+---
+
+# Medicine Quality Control Laboratories Dataset
+
+## Abstract
+
+**30,000 simulated QC laboratory assessments** (10,000 per scenario) across three laboratory tiers in sub-Saharan Africa. Variables include staff, equipment functionality, testing volume, turnaround time, proficiency testing, budget, reagent stockouts, and SF detection rates. Three scenarios: WHO-prequalified lab (20 staff, 92% equipment functional), national reference (8 staff, 65%), district basic (2 staff, 30%).
+
+**This dataset is entirely simulated. It must not be used for laboratory accreditation or regulatory decisions.**
+
+## Validation
+
+<p align="center">
+  <img src="validation_report.png" alt="Validation Report" width="100%">
+</p>
+
+## Usage
+
+```python
+from datasets import load_dataset
+dataset = load_dataset("electricsheepafrica/medicine-quality-control-labs", "national_reference_lab")
+df = dataset["train"].to_pandas()
+print(df[['staff_analysts', 'equipment_functional_pct', 'samples_tested_per_year']].describe())
+```
+
+## References
+
+1. WHO (2023). QC lab capacity in Africa.
+2. USP PQM+. Lab strengthening in 50+ countries.
+3. WHO Prequalification of QC Laboratories.
+4. African Medicines Quality Forum.
+
+## Citation
+
+```bibtex
+@dataset{esa_qc_labs_2025,
+  title   = {Medicine Quality Control Laboratories Dataset},
+  author  = {{Electric Sheep Africa}},
+  year    = {2025},
+  publisher = {Hugging Face},
+  url     = {https://huggingface.co/datasets/electricsheepafrica/medicine-quality-control-labs}
+}
+```
+
+## License
+
+[CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)

generate_dataset.py

@@ -0,0 +1,188 @@
+#!/usr/bin/env python3
+"""
+Literature-Informed Medicine Quality Control Laboratories Dataset
+==================================================================
+
+Each record = ONE QC laboratory assessment/performance record.
+
+Sources (v2.0):
+[1] WHO (2023). Only 30% of African countries have a functional
+    national QC lab. Many labs lack ISO 17025 accreditation.
+[2] USP PQM+. Supported 50+ labs globally. Equipment maintenance
+    and reagent supply are critical bottlenecks.
+[3] WHO Prequalification of QC Labs. Only ~10 WHO-prequalified
+    QC labs in Africa (out of ~60 globally).
+[4] African Medicines Quality Forum. Lab networking for
+    proficiency testing and harmonized methods.
+[5] Lancet Commission (2017). QC lab capacity directly impacts
+    SF medicine detection and regulatory enforcement.
+"""
+
+import numpy as np
+import pandas as pd
+import argparse
+import os
+
+EQUIPMENT_TYPES = [
+    'HPLC', 'UV_Vis_spectrophotometer', 'dissolution_apparatus',
+    'disintegration_tester', 'Karl_Fischer_titrator', 'IR_spectrophotometer',
+    'mass_spectrometer', 'GC', 'TLC_equipment', 'pH_meter',
+    'analytical_balance', 'stability_chamber',
+]
+
+TEST_TYPES = [
+    'identity', 'assay_API_content', 'dissolution', 'disintegration',
+    'uniformity_content', 'uniformity_mass', 'moisture_content',
+    'impurity_related_substances', 'microbial_limit', 'sterility',
+    'endotoxin', 'particle_size',
+]
+
+SCENARIOS = {
+    'who_prequalified_lab': {
+        'lab_tier': 'WHO_prequalified',
+        'ISO_17025': True,
+        'staff_analysts': 20,
+        'equipment_functional_rate': 0.92,
+        'samples_per_year': 5000,
+        'turnaround_days': 10,
+        'proficiency_testing_pass': 0.95,
+        'reagent_stockout_rate': 0.05,
+        'budget_usd': 800000,
+        'accreditation_scope_tests': 10,
+    },
+    'national_reference_lab': {
+        'lab_tier': 'national_reference',
+        'ISO_17025': False,
+        'staff_analysts': 8,
+        'equipment_functional_rate': 0.65,
+        'samples_per_year': 1500,
+        'turnaround_days': 25,
+        'proficiency_testing_pass': 0.70,
+        'reagent_stockout_rate': 0.25,
+        'budget_usd': 200000,
+        'accreditation_scope_tests': 5,
+    },
+    'district_basic_lab': {
+        'lab_tier': 'district_basic',
+        'ISO_17025': False,
+        'staff_analysts': 2,
+        'equipment_functional_rate': 0.30,
+        'samples_per_year': 200,
+        'turnaround_days': 45,
+        'proficiency_testing_pass': 0.35,
+        'reagent_stockout_rate': 0.55,
+        'budget_usd': 30000,
+        'accreditation_scope_tests': 2,
+    },
+}
+
+
+def generate_dataset(n=10000, seed=42, scenario='national_reference_lab'):
+    rng = np.random.default_rng(seed)
+    sc = SCENARIOS[scenario]
+    records = []
+
+    for idx in range(n):
+        rec = {'id': idx + 1}
+        rec['lab_tier'] = sc['lab_tier']
+        rec['lab_id'] = f"QCL_{rng.integers(1, 100):03d}"
+        rec['country_id'] = f"C_{rng.integers(1, 55):03d}"
+        rec['region_type'] = rng.choice(['urban', 'peri_urban', 'rural'],
+            p=[0.70, 0.20, 0.10] if scenario == 'who_prequalified_lab'
+            else ([0.50, 0.30, 0.20] if scenario == 'national_reference_lab'
+                  else [0.20, 0.30, 0.50]))
+
+        rec['ISO_17025_accredited'] = 1 if sc['ISO_17025'] else (
+            1 if rng.random() < 0.05 else 0)
+        rec['WHO_prequalified_lab'] = 1 if scenario == 'who_prequalified_lab' else 0
+        rec['staff_analysts'] = max(1, int(rng.poisson(sc['staff_analysts'])))
+        rec['staff_with_MSc_PhD'] = max(0, int(rec['staff_analysts'] * np.clip(
+            rng.normal(0.50 if scenario == 'who_prequalified_lab' else
+                       (0.25 if scenario == 'national_reference_lab' else 0.05), 0.10),
+            0, 0.80)))
+        rec['continuous_training_hours_per_year'] = max(0, int(rng.normal(
+            80 if scenario == 'who_prequalified_lab' else
+            (30 if scenario == 'national_reference_lab' else 5), 15)))
+
+        rec['annual_budget_usd'] = max(5000, int(rng.lognormal(
+            np.log(sc['budget_usd']), 0.3)))
+        rec['budget_from_government_pct'] = round(np.clip(
+            rng.normal(60 if scenario == 'who_prequalified_lab' else
+                       (40 if scenario == 'national_reference_lab' else 20), 15), 5, 95), 1)
+        rec['budget_from_donors_pct'] = round(100 - rec['budget_from_government_pct'] -
+            np.clip(rng.normal(10, 5), 0, 30), 1)
+
+        # Equipment
+        rec['total_equipment_items'] = max(3, int(rng.poisson(
+            len(EQUIPMENT_TYPES) if scenario == 'who_prequalified_lab' else
+            (8 if scenario == 'national_reference_lab' else 4))))
+        rec['equipment_functional_pct'] = round(np.clip(
+            rng.normal(sc['equipment_functional_rate'] * 100, 8), 10, 100), 1)
+        rec['HPLC_available'] = 1 if scenario != 'district_basic_lab' or rng.random() < 0.10 else 0
+        rec['dissolution_available'] = 1 if scenario == 'who_prequalified_lab' or (
+            scenario == 'national_reference_lab' and rng.random() < 0.60) else 0
+        rec['mass_spec_available'] = 1 if scenario == 'who_prequalified_lab' and rng.random() < 0.70 else 0
+
+        rec['reagent_stockout_days_per_year'] = max(0, int(rng.exponential(
+            sc['reagent_stockout_rate'] * 365 * 0.3)))
+        rec['reference_standards_available'] = 1 if rng.random() < (
+            0.95 if scenario == 'who_prequalified_lab' else
+            (0.55 if scenario == 'national_reference_lab' else 0.15)) else 0
+
+        # Performance
+        rec['samples_tested_per_year'] = max(10, int(rng.poisson(sc['samples_per_year'])))
+        rec['turnaround_days_median'] = max(1, int(rng.exponential(sc['turnaround_days'] * 0.6)))
+        rec['test_types_offered'] = min(len(TEST_TYPES), max(1, int(rng.poisson(
+            sc['accreditation_scope_tests']))))
+
+        rec['proficiency_testing_enrolled'] = 1 if rng.random() < (
+            0.98 if scenario == 'who_prequalified_lab' else
+            (0.50 if scenario == 'national_reference_lab' else 0.08)) else 0
+        rec['proficiency_testing_pass_rate'] = round(np.clip(
+            rng.normal(sc['proficiency_testing_pass'] * 100, 8), 0, 100), 1) if rec['proficiency_testing_enrolled'] else 0
+
+        rec['internal_audit_conducted'] = 1 if rng.random() < (
+            0.95 if scenario == 'who_prequalified_lab' else
+            (0.35 if scenario == 'national_reference_lab' else 0.05)) else 0
+        rec['CAPA_system_functional'] = 1 if rng.random() < (
+            0.90 if scenario == 'who_prequalified_lab' else
+            (0.25 if scenario == 'national_reference_lab' else 0.03)) else 0
+
+        rec['sf_detection_rate_pct'] = round(np.clip(
+            rng.normal(18 if scenario == 'who_prequalified_lab' else
+                       (22 if scenario == 'national_reference_lab' else 35), 5), 2, 60), 1)
+        rec['confirmatory_testing_capacity'] = 1 if scenario == 'who_prequalified_lab' or (
+            scenario == 'national_reference_lab' and rng.random() < 0.40) else 0
+
+        rec['year'] = rng.choice([2020, 2021, 2022, 2023, 2024],
+                                 p=[0.10, 0.15, 0.20, 0.25, 0.30])
+
+        records.append(rec)
+
+    df = pd.DataFrame(records)
+    print(f"\n{'='*65}")
+    print(f"QC Labs — {scenario} (n={n}, seed={seed})")
+    print(f"{'='*65}")
+    print(f"  Lab tier: {sc['lab_tier']}")
+    print(f"  Avg staff: {df['staff_analysts'].mean():.0f}")
+    print(f"  Equipment functional: {df['equipment_functional_pct'].mean():.0f}%")
+    print(f"  Samples/year: {df['samples_tested_per_year'].mean():.0f}")
+    print(f"  Proficiency pass: {df[df['proficiency_testing_enrolled']==1]['proficiency_testing_pass_rate'].mean():.0f}%")
+    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'qc_lab_{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', 'qc_lab_national_reference_lab.csv'), index=False)

requirements.txt

@@ -0,0 +1,3 @@
+numpy>=1.24
+pandas>=2.0
+matplotlib>=3.7

validate_dataset.py

@@ -0,0 +1,108 @@
+#!/usr/bin/env python3
+"""Validation & Diagnostic Visualization for Medicine Quality Control Laboratories Dataset."""
+
+import pandas as pd
+import numpy as np
+import matplotlib.pyplot as plt
+import os
+
+SCENARIOS = ['who_prequalified_lab', 'national_reference_lab', 'district_basic_lab']
+
+
+def load_scenarios(data_dir='data'):
+    dfs = {}
+    for sc in SCENARIOS:
+        path = os.path.join(data_dir, f'qc_lab_{sc}.csv')
+        if os.path.exists(path):
+            dfs[sc] = pd.read_csv(path)
+    return dfs
+
+
+def make_report(dfs, output='validation_report.png'):
+    fig, axes = plt.subplots(4, 2, figsize=(16, 24))
+    fig.suptitle(
+        'Medicine Quality Control Laboratories — Validation Report\n'
+        '(WHO-PQ Lab → National Reference → District Basic)',
+        fontsize=15, fontweight='bold', y=0.99)
+    colors = ['#2ecc71', '#f39c12', '#e74c3c']
+    x = np.arange(len(SCENARIOS))
+    labels = ['WHO-PQ', 'National Ref', 'District']
+
+    ax = axes[0, 0]
+    staff = [dfs[sc]['staff_analysts'].mean() for sc in SCENARIOS if sc in dfs]
+    ax.bar(x, staff, color=colors, alpha=0.8)
+    ax.set_xticks(x); ax.set_xticklabels(labels, fontsize=9)
+    for i, v in enumerate(staff):
+        ax.text(i, v+0.3, f'{v:.0f}', ha='center', fontsize=10, fontweight='bold')
+    ax.set_ylabel('Staff'); ax.set_title('Average Analyst Staff')
+
+    ax = axes[0, 1]
+    equip = [dfs[sc]['equipment_functional_pct'].mean() for sc in SCENARIOS if sc in dfs]
+    ax.bar(x, equip, color=colors, alpha=0.8)
+    ax.set_xticks(x); ax.set_xticklabels(labels, fontsize=9)
+    for i, v in enumerate(equip):
+        ax.text(i, v+1, f'{v:.0f}%', ha='center', fontsize=10, fontweight='bold')
+    ax.set_ylabel('Rate (%)'); ax.set_title('Equipment Functional Rate')
+
+    ax = axes[1, 0]
+    samp = [dfs[sc]['samples_tested_per_year'].mean() for sc in SCENARIOS if sc in dfs]
+    ax.bar(x, samp, color=colors, alpha=0.8)
+    ax.set_xticks(x); ax.set_xticklabels(labels, fontsize=9)
+    for i, v in enumerate(samp):
+        ax.text(i, v+30, f'{v:.0f}', ha='center', fontsize=10, fontweight='bold')
+    ax.set_ylabel('Samples/Year'); ax.set_title('Annual Testing Volume')
+
+    ax = axes[1, 1]
+    prof = []
+    for sc in SCENARIOS:
+        if sc in dfs:
+            enrolled = dfs[sc][dfs[sc]['proficiency_testing_enrolled']==1]
+            prof.append(enrolled['proficiency_testing_pass_rate'].mean() if len(enrolled) > 0 else 0)
+    ax.bar(x, prof, color=colors, alpha=0.8)
+    ax.set_xticks(x); ax.set_xticklabels(labels, fontsize=9)
+    for i, v in enumerate(prof):
+        ax.text(i, v+1, f'{v:.0f}%', ha='center', fontsize=10, fontweight='bold')
+    ax.set_ylabel('Pass Rate (%)'); ax.set_title('Proficiency Testing Pass Rate')
+
+    ax = axes[2, 0]
+    tat = [dfs[sc]['turnaround_days_median'].median() for sc in SCENARIOS if sc in dfs]
+    ax.bar(x, tat, color=colors, alpha=0.8)
+    ax.set_xticks(x); ax.set_xticklabels(labels, fontsize=9)
+    for i, v in enumerate(tat):
+        ax.text(i, v+0.5, f'{v:.0f}d', ha='center', fontsize=10, fontweight='bold')
+    ax.set_ylabel('Days'); ax.set_title('Median Turnaround Time')
+
+    ax = axes[2, 1]
+    budget = [dfs[sc]['annual_budget_usd'].mean()/1000 for sc in SCENARIOS if sc in dfs]
+    ax.bar(x, budget, color=colors, alpha=0.8)
+    ax.set_xticks(x); ax.set_xticklabels(labels, fontsize=9)
+    for i, v in enumerate(budget):
+        ax.text(i, v+5, f'${v:.0f}K', ha='center', fontsize=10, fontweight='bold')
+    ax.set_ylabel('Budget (USD K)'); ax.set_title('Average Annual Budget')
+
+    ax = axes[3, 0]
+    sf = [dfs[sc]['sf_detection_rate_pct'].mean() for sc in SCENARIOS if sc in dfs]
+    ax.bar(x, sf, color=colors, alpha=0.8)
+    ax.set_xticks(x); ax.set_xticklabels(labels, fontsize=9)
+    for i, v in enumerate(sf):
+        ax.text(i, v+0.5, f'{v:.0f}%', ha='center', fontsize=10, fontweight='bold')
+    ax.set_ylabel('SF Rate (%)'); ax.set_title('SF Detection Rate')
+
+    ax = axes[3, 1]
+    stockout = [dfs[sc]['reagent_stockout_days_per_year'].mean() for sc in SCENARIOS if sc in dfs]
+    ax.bar(x, stockout, color=colors, alpha=0.8)
+    ax.set_xticks(x); ax.set_xticklabels(labels, fontsize=9)
+    for i, v in enumerate(stockout):
+        ax.text(i, v+0.5, f'{v:.0f}d', ha='center', fontsize=10, fontweight='bold')
+    ax.set_ylabel('Days/Year'); ax.set_title('Reagent Stockout Days')
+
+    plt.tight_layout(rect=[0, 0, 1, 0.97])
+    plt.savefig(output, dpi=150, bbox_inches='tight')
+    print(f'Saved validation report to {output}')
+    plt.close()
+
+
+if __name__ == '__main__':
+    dfs = load_scenarios()
+    if dfs:
+        make_report(dfs)