---
license: cc-by-4.0
task_categories:
  - tabular-classification
language:
  - en
tags:
  - healthcare
  - epilepsy
  - neurology
  - seizure
  - antiepileptic-drug
  - treatment-gap
  - sub-saharan-africa
  - lmic
pretty_name: "Epilepsy & Neurological Disorders (Seizures, AED Access, Treatment Gap, Stigma)"
size_categories:
  - 10K<n<100K
configs:
  - config_name: neurology_clinic
    data_files: data/epilepsy_neurology_clinic.csv
  - config_name: district_hospital
    data_files: data/epilepsy_district_hospital.csv
    default: true
  - config_name: rural_health_centre
    data_files: data/epilepsy_rural_health_centre.csv
---

# Epilepsy & Neurological Disorders Dataset

## Abstract

This dataset provides **30,000 simulated epilepsy patient records** (10,000 per scenario) from sub-Saharan Africa. Each record contains 45+ variables including seizure type, etiology, AED access, treatment gap, stigma, psychosocial impact, and outcomes. Three settings: neurology clinic (25% treatment gap), district hospital (55%), and rural health centre (80%).

## 1. Introduction

Epilepsy prevalence in SSA is 4-15 per 1,000 — among the highest globally. The treatment gap exceeds 75% in rural areas. Phenobarbital is often the only AED available. There are fewer than 0.1 neurologists per 100,000 population. EEG is rarely available outside urban centres. Stigma drives 40-50% of patients to conceal their diagnosis and seek traditional/faith healers first.

**This dataset is entirely simulated. It must not be used for clinical decision-making.**

## 2. Methodology

### 2.1 Parameterization

| Parameter | Value | Source |
| --- | --- | --- |
| Prevalence SSA | 4-15/1000 | PMC 2024 |
| Treatment gap rural | >75% | PMC 2012 |
| Neurologists per 100K | <0.1 | Lancet Neurology |
| Phenobarbital only AED | Common | PMC 2024 |
| Stigma prevalence | 40-50% | Literature |

### 2.2 Scenario Design

| Scenario | Neurologist | EEG | AEDs | Treatment Gap |
| --- | --- | --- | --- | --- |
| Neurology clinic | Yes | Yes | Multiple | 25% |
| District hospital | No | No | Phenobarbital | 55% |
| Rural HC | No | No | Often stocked out | 80% |

## 3. Schema

| Column | Type | Description |
| --- | --- | --- |
| id | int | Unique identifier |
| age | int | Patient age |
| seizure_type | categorical | GTC / focal / absence / myoclonic |
| etiology | categorical | unknown / perinatal / CNS infection / TBI |
| in_treatment_gap | binary | Not receiving any AED |
| aed_prescribed | categorical | phenobarbital / carbamazepine / valproate / none |
| aed_adherent | binary | Taking AED as prescribed |
| stigma_experienced | binary | Social stigma |
| traditional_healer_consulted | binary | Consulted traditional healer |
| seizure_free_12m | binary | No seizures in 12 months |

## 4. Validation

<p align="center">
  <img src="validation_report.png" alt="Validation Report" width="100%">
</p>

Key validation checks:

- **Treatment gap**: 25% → 55% → 80% ✓
- **AED prescribed**: 76% → 36% → 16% ✓
- **Seizure-free 12m**: 15% → 8% → 3% ✓
- **Stigma**: ~50% across scenarios ✓
- **Traditional healer**: ~40% ✓
- **Stock-outs**: 10% → 25% → 45% ✓

## 5. Usage

```python
from datasets import load_dataset
dataset = load_dataset("electricsheepafrica/epilepsy-neurological", "district_hospital")
df = dataset["train"].to_pandas()
```

## 6. Limitations

- **Simulated**: Not from real epilepsy registries.
- **No EEG data**: No waveform or neuroimaging data.
- **Simplified**: No detailed pharmacokinetics.
- **No longitudinal**: Single time-point per patient.

## 7. References

1. PMC (2012). Epilepsy treatment SSA closing the gap.
2. PMC (2024). Epilepsy in Africa multifaceted perspective.
3. WHO (2024). Epilepsy fact sheet.
4. Lancet Neurology. Neurology workforce SSA.

## Citation

```bibtex
@dataset{esa_epilepsy_neuro_2025,
  title={Epilepsy and Neurological Disorders Dataset},
  author={Electric Sheep Africa},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/electricsheepafrica/epilepsy-neurological}
}
```

## License

[CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)