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

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.

Authors

Institutions

Publication Details

Journal
npj Digital Medicine
Published
2025-12-26
DOI
https://doi.org/10.1038/s41746-025-02277-8
Primary Topic
Artificial Intelligence in Healthcare and Education
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

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

Peng Yu, Lijuan Niu, MingYu Chen, Dongqiang Ye et al.
npj Digital Medicine
Artificial Intelligence in Healthcare and Education
article

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

Peng Yu, Lijuan Niu, MingYu Chen, Dongqiang Ye, W D Shen, Yongxin Wang, Chunbo Wang, Si-Xuan Liu, Hongyang Ma, Qianling Ye, Huaxia Yang, He Gao, Long Chen, Huimin Zhang, Rui Lin, Nan Liang, Jianxiong Wu, Yunlu Gao, Q M Zhao, Yunhui Tan, Lina Zhou, Kehang Mao, Yajie Ji, Yinan Jiang, Zhihui Tang, Shirui Wang, Lingmin Meng, Yue Liu, Mengran Lang, Zhicheng Huang, Wubin Sun, Tiantian Gu, Lingyun Ma, Qiuhong Gong, Zhihao Wang, Zeliang Lian, Haitao Shen, Youtao Yu
article en

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.

npj Digital MedicineVol. 9(1)
King University (US), Tongji University (CN), Xinjiang Medical University (CN), Harbin Medical University (CN), Jinan University (CN), Chinese Academy of Medical Sciences & Peking Union Medical College (CN), Peking University (CN), Fudan University (CN), Peking Union Medical College Hospital (CN), Chinese PLA General Hospital (CN), National Cancer Center (US), Shanghai University of Traditional Chinese Medicine (CN), Third Affiliated Hospital of Harbin Medical University (CN), EuroEspes Biomedical Research Center (ES), Beijing Anzhen Hospital (CN), People's Hospital of Xinjiang Uygur Autonomous Region (CN), Second Affiliated Hospital of Harbin Medical University (CN), Seventh People's Hospital of Shanghai (CN), Shuguang Hospital (CN), Intelligent Health (United Kingdom) (GB), Shanghai Skin Disease Hospital (CN), Shanxi Academy of Medical Sciences (CN), Huashan Hospital (CN), Tongji Hospital (CN), Beijing Haidian Hospital (CN)
Peace, Justice and strong institutions
Openalex Percentile: Top 39%
Artificial Intelligence in Healthcare and Education
DOIOpen Access PDFOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.