---
pretty_name: MedCase-Bench
size_categories:
- n<1K
tags:
- medical-literature
- case-reports
- clinical-atoms
configs:
- config_name: metadata
  default: true
  data_files:
  - split: evaluation
    path: metadata.json
---

# MedCase-Bench

[Project page](https://huggingface.co/spaces/PediaMedAI/MedCaseAgent)
· MedCase-150K
· [Agent code](https://github.com/PediaMedAI/MedCaseAgent)

MedCase-Bench contains **584 cases** for medical case-report generation, with
one clinical-atom JSON file per case and source bibliography. Clinical atoms
are patient-specific facts extracted from published case reports, covering
history, presentation, examinations, treatment and outcome.

Each atom file contains five lists of strings:

| Field | Contents |
| --- | --- |
| `history` | Patient background and relevant medical history. |
| `presentation` | Symptoms and findings at presentation. |
| `diagnostics` | Examinations, laboratory results and diagnostic observations. |
| `management` | Treatments, procedures and clinical decisions. |
| `outcome` | Clinical course, response and follow-up. |

## Files

| File | Contents |
| --- | --- |
| `atoms/<PMCID>_atoms.json` | 584 clinical-atom files. |
| `metadata.json` | 584 corresponding source records, including article identifiers, publication dates, licenses and source download links. |

This package contains textual inputs and source metadata. Full articles and
medical images can be obtained through the source links in `metadata.json`.
Source publications retain the licenses recorded in their metadata.

## Use

```python
import json
from pathlib import Path

files = sorted(Path("atoms").glob("*_atoms.json"))
assert len(files) == 584
atoms = json.loads(files[0].read_text(encoding="utf-8"))
print(atoms["history"])
```

After installing [MedCaseAgent](https://github.com/PediaMedAI/MedCaseAgent) and
configuring a compatible model API:

```bash
medcase-agent validate atoms/<PMCID>_atoms.json
medcase-agent generate atoms/<PMCID>_atoms.json \
  --exclude-ids path/to/your-exclusions.json --output runs
```

For generation with images, prepare each case's source images following the
[preprocessing guide](https://github.com/PediaMedAI/MedCaseAgent/blob/main/docs/preprocessing.md).
For retrieval-based evaluation, prepare your own PMCID, PMID and DOI exclusion
list for the benchmark sources and any related publications.