A large-scale dataset of structured clinical case narratives and question-answer pairs from case reports

This repository provides version v1.0.0 of a large-scale linked dataset of structured clinical case narratives and case-grounded question–answer pairs derived from full-text clinical case reports in the PubMed Central Open Access subset.

The dataset was constructed from PMC case-report articles published between 2000 and 2025. After license-aware filtering for textual redistribution, the release contains 76,028 license-eligible case-report articles, 101,965 structured case records, and 614,597 linked question–answer items. The resource is organized as a multi-level dataset spanning article-level source representations, case-level structured clinical narratives, intermediate task-planning records, and final QA items.

This dataset is intended to support non-commercial research and educational use in biomedical natural language processing, clinical narrative structuring, case-based retrieval, medical question answering, case-grounded reasoning, information extraction, and dataset engineering. It is not intended for clinical decision-making, direct patient care, patient profiling, or attempts to re-identify individuals described in the source case reports.

The dataset is released under the Creative Commons Attribution-NonCommercial 4.0 International license (CC BY-NC 4.0). Because the records are derived from PMC Open Access articles with heterogeneous source licenses, source-article provenance and license metadata are provided where available. Users are responsible for complying with both the dataset-level license and the license terms of the corresponding source articles. Textual derivative records from articles with no-derivatives, unknown, non-standard, or otherwise unclassified license terms were excluded from the redistributed textual release or retained only as non-copyrightable metadata where applicable.

Please cite the associated Data Descriptor and Article when using this dataset.