---
configs:
  - config_name: default
    data_files:
      - split: sample
        path: MedLayXPlain_sample.parquet
      - split: train
        path: MedLayXPlain_train.parquet
      - split: validation
        path: MedLayXPlain_val.parquet
      - split: test
        path: MedLayXPlain_test.parquet
      - split: excluded
        path: MedLayXPlain_excluded.parquet
license: cc-by-nc-sa-4.0
task_categories:
  - image-to-text
  - text-generation
language:
  - en
tags:
  - medical
  - radiology
  - pathology
  - lay-language
  - UMLS
  - vision-language
  - benchmark
size_categories:
  - 100K<n<1M
---

# MedLayXPlain

122,789 medical image samples with paired expert and lay captions, anchored in a hierarchical UMLS ontology (7 semantic groups, 43 TUIs, 2,411 CUIs). 8 imaging modalities. Each sample includes verification scores, sub-scores, and structured UMLS entity annotations.

**Splits**: sample (100 curated) / train (79,715) / validation (18,484) / test (24,590) / excluded (9,474)

```python
from datasets import load_dataset
ds = load_dataset("anonymous-medical/MedLayXPlain")
```

License: CC BY-NC-SA 4.0