NPCBench benchmark evaluates multimodal care-pathway reasoning for nasopharyngeal carcinoma
Original title: NPCBench: A Clinical Apprenticeship Benchmark for Multimodal Care-Pathway Reasoning in Nasopharyngeal Carcinoma
NPCBench includes 4,777 items and a 100-patient CarePath cohort with 417 scored tasks, revealing composition gaps where strong individual model performance does not ensure jointly correct diagnostic and care answers.
NPCBench is introduced as a clinical apprenticeship benchmark for nasopharyngeal carcinoma that mirrors specialist training from guideline knowledge to case planning, CarePath integration, and open-ended consultation. It contains 4,777 component and consultation items plus a 100-patient CarePath cohort. Evaluation across 30 models shows strong performance on isolated tasks but at most 23% strict CarePath success among 21 multimodal models, highlighting gaps in evidence grounding, longitudinal revision, and consultation judging.
- Contains 4,777 component items and 100-patient CarePath cohort.
- Tests whether diagnostic and care answers are jointly correct within pathways.
- 21 multimodal models reach at most 23% strict CarePath success.
- Benchmark evaluates grounding in visual evidence and clinical requirements.
- Paired controls examine evidence grounding and consultation judging.
From the original article
1 min read at huggingface.co
Current medical benchmarks typically break down clinical competence into isolated factual questions, single-image tasks, or static case vignettes. These tests score individual answers, but reveal less about whether a model can integrate subspecialty knowledge with multimodal evidence and track a patient's condition throughout a course of care.
By Pengkai Wang, Wei-Wei Zhang, Yan Li · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.