| --- |
| pretty_name: MultiCaRe Cases |
| license: cc-by-4.0 |
| task_categories: |
| - text-classification |
| - text-retrieval |
| language: |
| - en |
| size_categories: |
| - 100K<n<1M |
| --- |
| |
| # MultiCaRe: Open-Source Clinical Case Dataset |
|
|
| MultiCaRe is an open-source, multimodal clinical case dataset derived from PubMed Central’s Open Access (OA) Case Report articles. It links de-identified case narratives to figure images/captions and article-level metadata, enabling cross-modal supervision and retrieval. |
|
|
| - Source and process: OA case reports from PMC; parsed metadata and abstracts; extracted case narratives; downloaded and processed figures; aligned captions; curated image taxonomy (>140 classes). |
| - Scale: 85k+ articles with 110k+ patient mentions and 160k+ images (v2.0). |
| - Tasks enabled: narrative classification, retrieval, summarization; multimodal modeling with image joins; VQA/doc-QA with figure references. |
| - Citation: Paper — https://www.mdpi.com/2306-5729/10/8/123; Zenodo — https://zenodo.org/records/13936721. |
|
|
| This repository: per-case dataset |
| Per-case clinical narratives and demographics extracted from case reports. |
|
|
| Schema |
| - case_id: case identifier (joins to images.patient_id) |
| - article_id: PMCID (joins to articles.article_id) |
| - case_text: clinical case narrative |
| - age: age in years (0 if <1 y.o.) |
| - gender: Female, Male, Transgender, Unknown |
| |
| Quick start |
| ```python |
| from datasets import load_dataset |
| cases = load_dataset("openmed-community/multicare-cases", split="train") |
| print(cases[0]["case_text"][:600]) |
| ``` |
| |
| Join with images |
| ```python |
| from datasets import load_dataset |
| cases = load_dataset("openmed-community/multicare-cases", split="train") |
| imgs = load_dataset("openmed-community/multicare-images", split="train") |
|
|
| cid = cases[0]["case_id"] |
| imgs_for_case = imgs.filter(lambda e: e["patient_id"] == cid) |
| imgs_for_case[0]["image"].show() |
| ``` |
| |
| Notes |
| - No official splits; recommend patient/article-level splitting to avoid leakage. |
| - Per-item OA licenses are provided at the image level and via articles. |
| |
