Clinician use of language models diverges from how the models are evaluated

• 类型：arxiv
• 标识：2610.11069
• 链接：https://arxiv.org/abs/2610.11069
• 主分类：evaluation
• 形态：benchmark
• 被引：0
• 被引来源：Semantic Scholar
• S2被引：0
• 影响力被引：0
• TLDR：Analysis of queries sent by physicians, advanced practice providers and nurses in 35 specialties to an institutional assistant during an eight-month roll-out shows that benchmark scores say little about how clinical AI performs on most of the work it is actually given, and evaluation should be matched to real clinical use.
• 待LLM分类：否
• 标题中文：临床医生使用语言模型的方式与模型评测方式存在偏差
• TLDR中文：对某机构内部助手在八个月推广期内收到的、由 35 个科室的医生、高级实践护士和护士所发出查询的分析表明，基准分数几乎无法反映临床 AI 在其所承接的大多数实际任务上的表现，评测应与真实临床使用相匹配。
• 来源文件：
• /inbox/tom/_candidates/2026-10-13-rag-retrieval-reranking-candidates.json
• /inbox/tom/_candidates/2026-10-13-agent-rag-longcontext-candidates.json
• [S2 enrich]