---
license: cc-by-4.0
task_categories:
  - tabular-classification
  - tabular-regression
language:
  - en
tags:
  - healthcare
  - medicine-quality
  - regulatory
  - NRA
  - WHO-GBT
  - AMRH
  - pharmacovigilance
  - inspection
  - sub-saharan-africa
  - lmic
pretty_name: "Pharmaceutical Regulatory Capacity (NRA Maturity, GBT Scores, SF Prevalence)"
size_categories:
  - 10K<n<100K
configs:
  - config_name: ml3_ml4_advanced
    data_files: data/reg_capacity_ml3_ml4_advanced.csv
  - config_name: ml2_developing
    data_files: data/reg_capacity_ml2_developing.csv
    default: true
  - config_name: ml1_minimal
    data_files: data/reg_capacity_ml1_minimal.csv
---

# Pharmaceutical Regulatory Capacity Dataset

## Abstract

This dataset provides **30,000 simulated NRA capacity assessments** (10,000 per scenario) across three WHO Global Benchmarking Tool maturity levels in sub-Saharan Africa. Each record represents one regulatory authority assessment with 40+ variables including staff levels, budget, 9 regulatory function scores, inspection coverage, PMS failure rates, pharmacovigilance capacity, and estimated SF prevalence. Three scenarios: ML3/ML4 advanced (score 69/100, 8% SF), ML2 developing (score 34/100, 22% SF), ML1 minimal (score 11/100, 40% SF).

**This dataset is entirely simulated. It must not be used for regulatory assessment or policy decisions.**

## 1. Introduction

The WHO Global Benchmarking Tool assesses NRA maturity on a scale from ML1 (no formal system) to ML4 (operating at advanced level). Only 4 African NRAs have achieved ML3+ status (Tanzania TMDA, Ghana FDA, Nigeria NAFDAC, Egypt EDA). Regulatory capacity directly correlates with SF medicine prevalence — countries with ML1 NRAs face SF rates of 40%+ while ML3+ countries maintain rates below 10%.

## 2. Parameterization Evidence (v2.0)

| Parameter | Value Used | Source | Year |
| --- | --- | --- | --- |
| Only 4 African NRAs at ML3+ | 4 | WHO GBT | 2023 |
| 67% of LICs at ML1 | 67% | WHO landscape | 2022 |
| NRA staff range | 15-250 | WHO/AMRH reports | 2023 |
| SF correlation with regulation | Inverse | Lancet Commission | 2017 |
| AMRH initiative | AU/NEPAD | AMRH reports | 2023 |

## 3. Schema (key columns)

| Column | Type | Description |
| --- | --- | --- |
| maturity_level | categorical | ML1 / ML2 / ML3_ML4 |
| staff_total | int | Total NRA staff |
| budget_usd_millions | float | Annual budget |
| overall_regulatory_score | int | Composite score 0-100 |
| score_[function] | int | 9 regulatory function scores |
| inspection_coverage_pct | float | % of establishments inspected |
| PMS_failure_rate_pct | float | Post-market surveillance failure rate |
| sf_prevalence_estimated_pct | float | Estimated SF prevalence |
| pharmacovigilance_centre | binary | Has PV centre |
| WHO_listed_authority | binary | WHO-listed |
| lab_ISO_17025_accredited | binary | Lab accreditation |

## 4. Validation

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

## 5. Usage

```python
from datasets import load_dataset
dataset = load_dataset("electricsheepafrica/pharmaceutical-regulatory-capacity", "ml2_developing")
df = dataset["train"].to_pandas()
print(df[['overall_regulatory_score', 'sf_prevalence_estimated_pct']].describe())
```

## 6. Limitations

- **Simulated**: Not from real WHO GBT assessments.
- **Country-level aggregation**: Each record is an NRA snapshot, not granular.
- **Regulatory function scores independent**: In reality, functions are correlated.

## 7. References

1. WHO Global Benchmarking Tool. NRA maturity levels ML1-ML4.
2. AMRH/AU-NEPAD. African Medicines Regulatory Harmonization.
3. Lancet Commission (2017). Essential Medicines — regulatory capacity and SF prevalence.
4. WHO (2022). Global landscape of pharmaceutical regulation.

## Citation

```bibtex
@dataset{esa_reg_capacity_2025,
  title   = {Pharmaceutical Regulatory Capacity Dataset},
  author  = {{Electric Sheep Africa}},
  year    = {2025},
  publisher = {Hugging Face},
  url     = {https://huggingface.co/datasets/electricsheepafrica/pharmaceutical-regulatory-capacity}
}
```

## License

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