PubMed Central case reports provide 167,000 patient summaries extracted by author zhengyun21. The dataset also includes 3.1 million patient-article relevance scores and 293,000 patient-patient similarity pairs. A second version, PMC-Patients-V2, was released in December 2024 and contains 250,294 patients based on the 2024 PMC baseline.
Use Cases
- Train patient similarity models based on the 293,000 patient-patient similarity pairs mentioned in the description
- Develop patient-article relevance ranking systems based on the 3.1 million relevance scores mentioned in the description
- Generate synthetic patient summaries for medical education based on the extracted case report summaries
- Benchmark information extraction models for clinical text based on the patient summaries extracted from PubMed Central
Strengths
- Contains 167,000 patient summaries extracted from a large corpus of medical literature
- Includes 3.1 million patient-article relevance annotations
- Provides 293,000 patient-patient similarity pairs for comparative analysis
Limitations
- Column-level documentation is absent; field semantics must be inferred after download
- Row count for the original version is unknown, which may limit suitability assessment
- Description metadata is limited; actual data quality requires manual inspection after download
Provenance
- Source
- PubMed Central (PMC)
- Collection Method
- Extracted from case reports in PMC articles
- Freshness
- Last updated 2024-12-10 09:21:26; freshness should be verified