Upload folder using huggingface_hub

README.md

@@ -0,0 +1,92 @@
+---
+license: cc-by-4.0
+task_categories:
+  - tabular-classification
+language:
+  - en
+tags:
+  - environmental-health
+  - asbestos
+  - mesothelioma
+  - occupational-health
+  - lung-cancer
+  - synthetic
+  - sub-saharan-africa
+pretty_name: Asbestos Exposure & Mesothelioma (SSA)
+size_categories:
+  - 10K<n<100K
+configs:
+  - config_name: former_mining_community
+    data_files: data/asbestos_mining_community.csv
+    default: true
+  - config_name: urban_construction
+    data_files: data/asbestos_urban_construction.csv
+  - config_name: rural_asbestos_roofing
+    data_files: data/asbestos_rural_roofing.csv
+---
+
+# Asbestos Exposure & Mesothelioma in Sub-Saharan Africa
+
+## Abstract
+
+Synthetic dataset modelling asbestos exposure pathways, fibre types, and health outcomes (mesothelioma, asbestosis, lung cancer) across three settings in SSA. South Africa was a global leader in asbestos production; Wagner (1960) discovered the mesothelioma link there. WHO Africa reports asbestos use continues despite warnings, particularly in roofing, construction, and brake linings. Latency period is 20-50 years.
+
+### Scenarios
+
+- **Former Mining Community**: South Africa-type communities near closed asbestos mines with high crocidolite/amosite exposure.
+- **Urban Construction**: Cities with ongoing chrysotile use in building materials and demolition.
+- **Rural Asbestos Roofing**: Widespread asbestos-cement roofing in rural areas with chronic low-level exposure.
+
+## Parameterization Evidence
+
+| Parameter | Value | Source | Year |
+| --- | --- | --- | --- |
+| Asbestos causes mesothelioma, asbestosis, lung cancer | Health effects | WHO Fact Sheet; IARC Group 1 | 2023 |
+| SA global leader in asbestos production; ban in 2002 | History | ScienceDirect; asbestos.com | 2004 |
+| Wagner (1960) discovered mesothelioma-asbestos link in SA | Discovery | PMC1522094 | 2005 |
+| Asbestos use continues in Africa despite warnings | Ongoing use | WHO Africa | 2023 |
+| Eastern SSA: substantial increases in asbestos lung cancer | GBD trend | PMC12573932 (GBD 2021) | 2024 |
+| Mesothelioma mortality lower than expected in SA due to HIV | Co-morbidity | PubMed 21422006 | 2011 |
+| Latency period 20-50 years | Disease natural history | WHO | 2023 |
+
+## Validation
+
+![Validation Report](validation_report.png)
+
+## Usage
+
+```python
+from datasets import load_dataset
+ds = load_dataset("electricsheepafrica/asbestos-mesothelioma", "former_mining_community")
+```
+
+## Limitations
+
+- Synthetic data; not for clinical decision-making.
+- Latency modelling simplified; real exposure-disease relationships are complex.
+- Does not capture legacy contamination mapping or remediation status.
+
+## References
+
+1. WHO. Asbestos Fact Sheet. 2023.
+2. WHO Africa. Asbestos use continues in Africa. 2023.
+3. PMC1522094. Asbestos-related disease in South Africa. 2005.
+4. Wagner JC. Diffuse pleural mesothelioma and asbestos exposure in SA. *Br J Ind Med*, 1960.
+5. PMC12573932. Global burden of lung cancer from occupational asbestos (GBD 2021). 2024.
+6. PubMed 21422006. Mesothelioma mortality trends in South Africa 1995-2007. 2011.
+
+## Citation
+
+```bibtex
+@dataset{electricsheepafrica_asbestos_mesothelioma_2025,
+  title={Asbestos Exposure and Mesothelioma in Sub-Saharan Africa},
+  author={Electric Sheep Africa},
+  year={2025},
+  publisher={HuggingFace},
+  url={https://huggingface.co/datasets/electricsheepafrica/asbestos-mesothelioma}
+}
+```
+
+## License
+
+CC-BY-4.0

generate_dataset.py

@@ -0,0 +1,223 @@
+"""Generate synthetic asbestos exposure & mesothelioma dataset for SSA.
+
+Research-based parameterization:
+- WHO Africa: Asbestos use continues despite warnings; used in roofing,
+  insulation, cement pipes, brake linings across SSA.
+- WHO Fact Sheet: Asbestos causes lung/larynx/ovary cancer, mesothelioma,
+  asbestosis. All forms carcinogenic (IARC Group 1).
+- South Africa: Global leader in asbestos production; crocidolite/amosite/
+  chrysotile all mined. Wagner (1960) discovered mesothelioma link.
+- PMC1522094: Social production of asbestos-related disease in SA;
+  asbestosis, lung cancer, mesothelioma since early 1900s.
+- PMC12573932 (GBD 2021): Eastern SSA saw substantial increases in lung
+  cancer from occupational asbestos exposure.
+- SA banned asbestos mining in 2002; many SSA countries still use.
+- Latency period: 20-50 years from exposure to mesothelioma.
+- Mesothelioma mortality rates lower than expected in SA due to HIV
+  reducing life expectancy (PubMed 21422006).
+"""
+
+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 = {
+    # Former mining communities (South Africa type)
+    "former_mining_community": {
+        "setting_probs": {"rural_mining": 0.50, "peri_urban": 0.30, "urban": 0.20},
+        "exposure_probs": {"mining_direct": 0.30, "mining_environmental": 0.25,
+                           "construction": 0.15, "roofing_materials": 0.15,
+                           "household_exposure": 0.10, "brake_lining": 0.05},
+        "fibre_type_probs": {"crocidolite": 0.35, "amosite": 0.25, "chrysotile": 0.30, "mixed": 0.10},
+        "exposure_intensity_mean": 3.5,  # fibres/mL
+        "exposure_years_mean": 15,
+        "mesothelioma_rate": 0.008,
+        "asbestosis_prev": 0.12,
+        "lung_cancer_rate": 0.005,
+        "ban_in_place": 0.70,
+        "medical_surveillance": 0.15,
+    },
+    # Urban construction/demolition (ongoing use)
+    "urban_construction": {
+        "setting_probs": {"urban": 0.45, "peri_urban": 0.35, "industrial": 0.20},
+        "exposure_probs": {"construction": 0.30, "roofing_materials": 0.25,
+                           "demolition": 0.15, "insulation": 0.10,
+                           "household_exposure": 0.10, "brake_lining": 0.10},
+        "fibre_type_probs": {"chrysotile": 0.55, "amosite": 0.15, "crocidolite": 0.10, "mixed": 0.20},
+        "exposure_intensity_mean": 1.5,
+        "exposure_years_mean": 10,
+        "mesothelioma_rate": 0.003,
+        "asbestosis_prev": 0.06,
+        "lung_cancer_rate": 0.003,
+        "ban_in_place": 0.30,
+        "medical_surveillance": 0.05,
+    },
+    # Rural asbestos roofing communities
+    "rural_asbestos_roofing": {
+        "setting_probs": {"rural": 0.55, "peri_urban": 0.30, "urban": 0.15},
+        "exposure_probs": {"roofing_materials": 0.40, "household_exposure": 0.25,
+                           "water_pipes": 0.10, "construction": 0.10,
+                           "environmental": 0.10, "brake_lining": 0.05},
+        "fibre_type_probs": {"chrysotile": 0.60, "mixed": 0.20, "amosite": 0.10, "crocidolite": 0.10},
+        "exposure_intensity_mean": 0.5,
+        "exposure_years_mean": 20,
+        "mesothelioma_rate": 0.002,
+        "asbestosis_prev": 0.03,
+        "lung_cancer_rate": 0.002,
+        "ban_in_place": 0.15,
+        "medical_surveillance": 0.02,
+    },
+}
+
+SCENARIO_FILES = {
+    "former_mining_community": "asbestos_mining_community.csv",
+    "urban_construction": "asbestos_urban_construction.csv",
+    "rural_asbestos_roofing": "asbestos_rural_roofing.csv",
+}
+
+
+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"])
+        age = int(np.clip(rng.normal(45, 15), 18, 80))
+        sex = rng.choice(["male", "female"], p=[0.65, 0.35])
+
+        exposure_type = _choice(rng, params["exposure_probs"])
+        fibre_type = _choice(rng, params["fibre_type_probs"])
+        is_occupational = int(exposure_type in ("mining_direct", "construction", "demolition", "brake_lining"))
+        is_environmental = int(exposure_type in ("mining_environmental", "household_exposure",
+                                                  "roofing_materials", "environmental", "water_pipes"))
+
+        exposure_years = int(np.clip(
+            rng.normal(params["exposure_years_mean"], 8), 0, 45))
+        exposure_intensity = float(np.clip(
+            rng.lognormal(np.log(max(params["exposure_intensity_mean"], 0.1)), 0.8),
+            0.01, 50))
+        if not is_occupational:
+            exposure_intensity *= 0.2
+
+        cumulative_exposure = float(exposure_intensity * exposure_years)
+        latency_years = int(np.clip(rng.normal(30, 10), 10, 50))
+        time_since_first_exposure = int(np.clip(rng.normal(20, 10), 0, 50))
+
+        ppe_use = int(is_occupational and rng.random() < 0.10)
+        if ppe_use:
+            exposure_intensity *= 0.3
+
+        # Fibre potency (crocidolite > amosite > chrysotile)
+        potency = {"crocidolite": 2.0, "amosite": 1.5, "chrysotile": 1.0, "mixed": 1.3}
+        risk_mult = cumulative_exposure * potency.get(fibre_type, 1.0) / 20
+
+        # Health outcomes
+        # Mesothelioma (latency 20-50 yrs; crocidolite highest risk)
+        mesothelioma = int(time_since_first_exposure >= 15 and rng.random() < np.clip(
+            params["mesothelioma_rate"] * risk_mult, 0, 0.05))
+        mesothelioma_type = rng.choice(["pleural", "peritoneal"], p=[0.85, 0.15]) if mesothelioma else "none"
+
+        # Asbestosis (PMC1522094: progressive fibrotic lung disease)
+        asbestosis = int(exposure_years >= 5 and rng.random() < np.clip(
+            params["asbestosis_prev"] * risk_mult, 0, 0.30))
+
+        # Lung cancer
+        lung_cancer = int(age >= 40 and rng.random() < np.clip(
+            params["lung_cancer_rate"] * risk_mult, 0, 0.03))
+        smoking = int(rng.random() < 0.15)
+        if smoking:
+            lung_cancer = int(rng.random() < np.clip(
+                params["lung_cancer_rate"] * risk_mult * 5, 0, 0.10))  # synergy
+
+        # Pleural plaques (early marker)
+        pleural_plaques = int(exposure_years >= 10 and rng.random() < np.clip(
+            0.10 * risk_mult, 0, 0.40))
+        pleural_effusion = int(pleural_plaques and rng.random() < 0.10)
+
+        # Respiratory symptoms
+        dyspnoea = int(rng.random() < np.clip(0.10 + risk_mult * 0.05, 0, 0.35))
+        cough_chronic = int(rng.random() < np.clip(0.08 + risk_mult * 0.04, 0, 0.30))
+        reduced_fvc = int(asbestosis or rng.random() < np.clip(risk_mult * 0.03, 0, 0.15))
+
+        any_asbestos_disease = int(mesothelioma or asbestosis or lung_cancer or pleural_plaques)
+
+        # Compensation & regulation
+        ban_in_place = int(rng.random() < params["ban_in_place"])
+        medical_surveillance = int(rng.random() < params["medical_surveillance"])
+        compensation_claimed = int(any_asbestos_disease and rng.random() < 0.05)
+        chest_xray_done = int(rng.random() < 0.10)
+
+        # HIV co-morbidity (SA context: reduces life expectancy)
+        hiv_positive = int(rng.random() < 0.12)
+        died = int((mesothelioma and rng.random() < 0.85) or
+                   (lung_cancer and rng.random() < 0.70))
+
+        record = {
+            "record_id": f"{name[:3].upper()}-{idx:05d}",
+            "scenario": name,
+            "year": year,
+            "setting": setting,
+            "age": age,
+            "sex": sex,
+            "exposure_type": exposure_type,
+            "fibre_type": fibre_type,
+            "is_occupational": is_occupational,
+            "is_environmental": is_environmental,
+            "exposure_years": exposure_years,
+            "exposure_intensity_f_mL": round(exposure_intensity, 2),
+            "cumulative_exposure": round(cumulative_exposure, 1),
+            "latency_years": latency_years,
+            "time_since_first_exposure": time_since_first_exposure,
+            "ppe_use": ppe_use,
+            "smoking": smoking,
+            "mesothelioma": mesothelioma,
+            "mesothelioma_type": mesothelioma_type,
+            "asbestosis": asbestosis,
+            "lung_cancer": lung_cancer,
+            "pleural_plaques": pleural_plaques,
+            "pleural_effusion": pleural_effusion,
+            "dyspnoea": dyspnoea,
+            "cough_chronic": cough_chronic,
+            "reduced_fvc": reduced_fvc,
+            "any_asbestos_disease": any_asbestos_disease,
+            "ban_in_place": ban_in_place,
+            "medical_surveillance": medical_surveillance,
+            "compensation_claimed": compensation_claimed,
+            "chest_xray_done": chest_xray_done,
+            "hiv_positive": hiv_positive,
+            "died": died,
+        }
+        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()

requirements.txt

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

validate_dataset.py

@@ -0,0 +1,90 @@
+"""Validate synthetic asbestos exposure & mesothelioma dataset."""
+
+from __future__ import annotations
+
+from pathlib import Path
+
+import matplotlib.pyplot as plt
+import pandas as pd
+
+SCENARIO_FILES = {
+    "former_mining_community": "asbestos_mining_community.csv",
+    "urban_construction": "asbestos_urban_construction.csv",
+    "rural_asbestos_roofing": "asbestos_rural_roofing.csv",
+}
+
+COLORS = {"former_mining_community": "#e6550d", "urban_construction": "#756bb1", "rural_asbestos_roofing": "#31a354"}
+
+
+def load_data() -> pd.DataFrame:
+    frames = []
+    for scenario, filename in SCENARIO_FILES.items():
+        df = pd.read_csv(Path("data") / filename)
+        frames.append(df)
+    return pd.concat(frames, ignore_index=True)
+
+
+def plot_validation(df: pd.DataFrame, output_path: Path) -> None:
+    fig, axes = plt.subplots(4, 2, figsize=(14, 16))
+    axes = axes.flatten()
+
+    for s in SCENARIO_FILES:
+        subset = df[df["scenario"] == s]
+        axes[0].hist(subset["cumulative_exposure"], bins=40, alpha=0.5, color=COLORS[s], label=s, range=(0, 100))
+    axes[0].set_title("Cumulative Exposure Distribution (f/mL·yr)")
+    axes[0].legend(fontsize=7)
+
+    disease_cols = ["mesothelioma", "asbestosis", "lung_cancer", "pleural_plaques"]
+    dis = df.groupby("scenario")[disease_cols].mean() * 100
+    dis.plot(kind="bar", ax=axes[1])
+    axes[1].set_title("Asbestos-Related Disease (%)")
+    axes[1].legend(fontsize=6)
+
+    ft = df.groupby(["scenario", "fibre_type"]).size().groupby(level=0).apply(lambda s: s / s.sum())
+    ft.unstack().plot(kind="bar", stacked=True, ax=axes[2])
+    axes[2].set_title("Fibre Type Distribution")
+    axes[2].legend(fontsize=7)
+
+    exp = df.groupby(["scenario", "exposure_type"]).size().groupby(level=0).apply(lambda s: s / s.sum())
+    exp.unstack().plot(kind="bar", stacked=True, ax=axes[3])
+    axes[3].set_title("Exposure Type Distribution")
+    axes[3].legend(fontsize=5)
+
+    for s in SCENARIO_FILES:
+        subset = df[df["scenario"] == s]
+        axes[4].scatter(subset["cumulative_exposure"], subset["any_asbestos_disease"],
+                        s=4, alpha=0.05, color=COLORS[s], label=s)
+    axes[4].set_title("Cumulative Exposure vs Disease")
+    axes[4].legend(fontsize=7)
+
+    resp_cols = ["dyspnoea", "cough_chronic", "reduced_fvc"]
+    resp = df.groupby("scenario")[resp_cols].mean() * 100
+    resp.plot(kind="bar", ax=axes[5])
+    axes[5].set_title("Respiratory Symptoms (%)")
+    axes[5].legend(fontsize=7)
+
+    reg_cols = ["ban_in_place", "medical_surveillance", "ppe_use", "chest_xray_done"]
+    reg = df.groupby("scenario")[reg_cols].mean() * 100
+    reg.plot(kind="bar", ax=axes[6])
+    axes[6].set_title("Regulation & Surveillance (%)")
+    axes[6].legend(fontsize=6)
+
+    mort_cols = ["died", "hiv_positive", "compensation_claimed"]
+    mort = df.groupby("scenario")[mort_cols].mean() * 100
+    mort.plot(kind="bar", ax=axes[7])
+    axes[7].set_title("Mortality, HIV & Compensation (%)")
+    axes[7].legend(fontsize=7)
+
+    plt.tight_layout()
+    fig.savefig(output_path, dpi=200)
+    plt.close(fig)
+
+
+def main() -> None:
+    df = load_data()
+    plot_validation(df, Path("validation_report.png"))
+    print("Saved validation_report.png")
+
+
+if __name__ == "__main__":
+    main()