---
license: cc-by-4.0
task_categories:
  - tabular-classification
language:
  - en
tags:
  - healthcare
  - stroke
  - cerebrovascular
  - hypertension
  - neurology
  - thrombolysis
  - sub-saharan-africa
  - lmic
pretty_name: "Stroke & Cerebrovascular Disease (Type, Severity, Treatment, Disability)"
size_categories:
  - 10K<n<100K
configs:
  - config_name: tertiary_hospital
    data_files: data/stroke_tertiary_hospital.csv
  - config_name: district_hospital
    data_files: data/stroke_district_hospital.csv
    default: true
  - config_name: rural_facility
    data_files: data/stroke_rural_facility.csv
---

# Stroke & Cerebrovascular Disease Dataset

## Abstract

This dataset provides **30,000 simulated stroke records** (10,000 per scenario) of adults presenting with acute stroke in sub-Saharan Africa. Each record contains 50+ variables including risk factors, stroke type, severity (NIHSS, GCS), imaging, treatment, complications, and outcomes (30-day mortality, mRS). Three settings: tertiary hospital (28% mortality), district hospital (41%), and rural facility (57%).

## 1. Introduction

Stroke in Africa has an annual incidence up to 316/100K person-years, with 30-day case fatality rates of ~30% (range 18-40%) and 90-day mortality reaching 50% at some centres. Approximately 60% of strokes are ischaemic and 32% haemorrhagic — a higher proportion of haemorrhagic stroke than in HICs. Mean age at onset is ~59 years, 10 years younger than Western populations. Hypertension is the dominant risk factor with BP control rates <10% in many SSA populations. CT scanning and thrombolysis are virtually unavailable outside tertiary centres.

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

## 2. Methodology

### 2.1 Parameterization

| Parameter | Value | Source |
| --- | --- | --- |
| Incidence (Africa) | Up to 316/100K | Frontiers Stroke 2025 |
| 30-day CFR (SSA pooled) | ~30% | PLOS Glob PH 2024 |
| 90-day mortality | ~50% | Muhimbili 2018 |
| Ischaemic / ICH / SAH | 60% / 32% / 8% | Sierra Leone 2021 |
| Mean age | 59 years | Sierra Leone 2021 |
| Hypertension prevalence | 75% | PMC 2015 |
| BP control rate | <10% | PMC 2015 |
| Recurrence rate | 9-25% | PMC 2024 |

### 2.2 Scenario Design

| Scenario | CT | Thrombolysis | Stroke Unit | 30-Day Mortality |
| --- | --- | --- | --- | --- |
| Tertiary hospital | Yes | Limited | Yes | 28% |
| District hospital | No | No | No | 41% |
| Rural facility | No | No | No | 57% |

## 3. Schema

| Column | Type | Description |
| --- | --- | --- |
| id | int | Unique identifier |
| age_years | int | Patient age |
| sex | categorical | M / F |
| hypertension | binary | Hypertension |
| systolic_bp | int | Systolic BP (mmHg) |
| diastolic_bp | int | Diastolic BP (mmHg) |
| diabetes | binary | Diabetes |
| atrial_fibrillation | binary | AF |
| smoking | binary | Current smoker |
| alcohol_heavy | binary | Heavy alcohol |
| prior_stroke | binary | Prior stroke |
| stroke_type | categorical | ischaemic / intracerebral_haemorrhage / subarachnoid_haemorrhage |
| ischaemic_subtype | categorical | large_vessel / small_vessel / cardioembolic / undetermined |
| nihss_score | int | NIHSS (0-42) |
| gcs_score | int | GCS (3-15) |
| hemiparesis | binary | Hemiparesis |
| aphasia | binary | Aphasia |
| dysphagia | binary | Dysphagia |
| onset_to_arrival_hours | int | Time to presentation |
| ct_performed | binary | CT scan done |
| thrombolysis | binary | Thrombolysis given |
| antiplatelet | binary | Antiplatelet |
| statin | binary | Statin |
| physiotherapy_started | binary | Early physiotherapy |
| aspiration_pneumonia | binary | Aspiration pneumonia |
| outcome_30_day | categorical | survived / died |
| mrs_discharge | int | Modified Rankin Scale (0-5) |
| disability | categorical | none / mild / moderate / severe |
| hospital_days | int | Hospital stay |

## 4. Validation

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

Key validation checks:

- **Mortality gradient**: 28% → 41% → 57% ✓
- **Stroke type**: 60% ischaemic, 32% ICH ✓
- **Hypertension**: 75% prevalence ✓
- **CT access**: 80% tertiary vs 0% district/rural ✓
- **ICH mortality > ischaemic** ✓
- **Mean age ~59** ✓

## 5. Usage

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

## 6. Limitations

- **Simulated**: Not from real stroke registries.
- **No neuroimaging**: No actual CT/MRI data.
- **No longitudinal**: 30-day outcome only, no long-term follow-up.
- **No rehabilitation**: Limited rehab outcome data.
- **Simplified**: No detailed vascular territory mapping.

## 7. References

1. Frontiers Stroke (2025). Strategic action plan for stroke in Africa.
2. PLOS Glob Public Health (2024). 30-day stroke CFR in SSA.
3. PubMed (2021). Stroke register Sierra Leone.
4. PubMed (2020). Predictors of stroke mortality Uganda.
5. PubMed (2018). Stroke at Muhimbili Tanzania.
6. PMC (2015). Burden of stroke in Africa.
7. PMC (2024). Stroke recurrence in SSA.

## Citation

```bibtex
@dataset{esa_stroke_2025,
  title={Stroke and Cerebrovascular Disease Dataset},
  author={Electric Sheep Africa},
  year={2025},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/electricsheepafrica/stroke-cerebrovascular-disease}
}
```

## License

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