---
license: apache-2.0
language:
- en
tags:
- medical
- clinical-nlp
- plain-language
- summarization
size_categories:
- 1K<n<10K
---

# Mediclear Dataset (5000 rows)

Synthetic paired English dataset for **clinical shorthand → patient-friendly explanation**.

Derived from the schema and style of [`thenomdevel/mediclear_dataset_1000`](https://huggingface.co/datasets/thenomdevel/mediclear_dataset_1000).

## Schema

| Column | Description |
|--------|-------------|
| `id` | Row index (1–5000) |
| `source` | Clinical / abbreviated text |
| `target` | Plain-language explanation |
| `specialty` | One of 20 medical specialties |
| `document_type` | One of 9 note/report types |
| `complexity` | `high`, `medium`, or `low` |

## Files

- `mediclear_dataset_5000.csv` — UTF-8 CSV with header
- `mediclear_dataset_5000.jsonl` — JSON Lines (same fields)

## Generation

Regenerate locally:

```bash
python3 gen_5k.py
```

## Disclaimer

Synthetic educational data — **not** for clinical decision-making.