A novel evaluation benchmark for medical LLMs illuminating safety and effectiveness in clinical domains

TL;DR

A new benchmark called CSEDB evaluates medical LLMs on safety and effectiveness, revealing moderate performance and significant drops in high-risk scenarios.

Contribution

The novel CSEDB benchmark introduces 30 metrics based on clinical expert consensus for evaluating medical LLMs.

Findings

Medical LLMs showed moderate average performance (57.2%) with a 13.3% drop in high-risk scenarios.

Domain-specific models outperformed general-purpose models in safety and effectiveness metrics.

Abstract

Large language models (LLMs) hold promise in clinical decision support but face major challenges in safety evaluation and effectiveness validation. We developed the Clinical Safety-Effectiveness Dual-Track Benchmark (CSEDB), a multidimensional framework built on clinical expert consensus, encompassing 30 metrics covering critical areas like critical illness recognition, guideline adherence, and medication safety, with weighted consequence measures. Thirty-two specialist physicians developed and revised 2069 open-ended Q&A items aligned with these criteria, spanning 26 clinical departments to simulate real-world scenarios. Benchmark testing of six LLMs revealed moderate overall performance (average total score 57.2%, safety 54.7%, effectiveness 62.3%), with a significant 13.3% performance drop in high-risk scenarios (p < 0.0001). Domain-specific medical LLMs showed consistent performance advantages over general-purpose models, with relatively higher top scores in safety (0.912) and effectiveness (0.861). The findings of this study not only provide a standardized metric for evaluating the clinical application of medical LLMs, facilitating comparative analyses, risk exposure identification, and improvement directions across different scenarios, but also hold the potential to promote safer and more effective deployment of large language models in healthcare environments.

Figures