CUPCase: Clinically Uncommon Patient Cases and Diagnoses Dataset

CUPCase is a medical benchmark dataset for developing Large Language Models, presented at AAAI 2025. It focuses on clinically uncommon patient cases and diagnoses, aiming to address the complexity lacking in benchmarks derived from exam questions or medical literature. The dataset was created by ofir408 and last updated on April 16, 2025.

Use Cases

• Benchmarking LLM performance on medical knowledge extraction based on uncommon clinical cases.
• Training diagnostic models on complex, real-world patient scenarios described in the dataset.
• Evaluating model summarization capabilities for medical case narratives.

Strengths

• Dataset is associated with a peer-reviewed conference (AAAI 2025).
• Focuses on a specific niche of clinically uncommon cases, which may address a gap in existing benchmarks.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and license are unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Collection Method

Likely compiled from medical sources to address benchmark limitations, though the specific method is not detailed.

Time Range

null

Freshness

Last updated 2025-04-16 18:34:22; freshness should be verified.

Geography

null