MedArena：比较医学领域临床医生偏好的大型语言模型

Large language models (LLMs) are becoming increasingly integral to clinician workflows, encompassing clinical decision support, medical education, and patient communication. Current evaluation methods for medical LLMs predominantly rely on static, templated benchmarks that fail to adequately capture the complexity and dynamics of real-world clinical practice, leading to a disconnect between benchmark performance and clinical utility. To address these limitations, MedArena is introduced as an interactive evaluation platform that allows clinicians to directly test and compare leading LLMs using their own medical queries. Upon receiving a clinician-provided query, MedArena presents responses from two randomly selected models and prompts the user to select the preferred response. This platform aims to enhance the applicability of LLMs in actual clinical settings.