医疗多模态大语言模型的视点临床意图理解能力基准测试
Benchmarking Egocentric Clinical Intent Understanding Capability for Medical Multimodal Large Language Models
- Shenzhen University（深圳大学）
- Stanford University（斯坦福大学）
- City University of Hong Kong（香港城市大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本文提出MedGaze-Bench，首个评估医疗多模态大语言模型视点临床意图理解能力的基准测试，通过三维意图框架和陷阱QA机制，揭示现有模型在手术、急救和诊断任务中对意图理解的不足。
AI中文摘要：
医疗多模态大语言模型（Med-MLLMs）需要视点临床意图理解能力以实现实际应用，但现有基准测试未能评估这一关键能力。为解决这些挑战，我们引入了MedGaze-Bench，这是首个利用临床人员目光作为认知光标来评估手术、急救模拟和诊断解释中意图理解的基准测试。我们的基准测试解决了三个根本性挑战：解剖结构的视觉同质性、临床工作流程中的严格时间因果依赖性以及隐含的安全规程遵守。我们提出了一种三维临床意图框架，评估：（1）空间意图：在视觉噪声中区分精确目标，（2）时间意图：通过回顾性和前瞻性推理推断因果逻辑，（3）标准意图：通过安全检查验证规程合规性。除了准确性指标外，我们引入了陷阱QA机制，通过惩罚幻觉和认知趋同来压力测试临床可靠性。实验发现当前MLLMs在视点意图上存在问题，主要是过度依赖全局特征，导致生成虚假观察和无批判性接受无效指令。
英文摘要：
Medical Multimodal Large Language Models (Med-MLLMs) require egocentric clinical intent understanding for real-world deployment, yet existing benchmarks fail to evaluate this critical capability. To address these challenges, we introduce MedGaze-Bench, the first benchmark leveraging clinician gaze as a Cognitive Cursor to assess intent understanding across surgery, emergency simulation, and diagnostic interpretation. Our benchmark addresses three fundamental challenges: visual homogeneity of anatomical structures, strict temporal-causal dependencies in clinical workflows, and implicit adherence to safety protocols. We propose a Three-Dimensional Clinical Intent Framework evaluating: (1) Spatial Intent: discriminating precise targets amid visual noise, (2) Temporal Intent: inferring causal rationale through retrospective and prospective reasoning, and (3) Standard Intent: verifying protocol compliance through safety checks. Beyond accuracy metrics, we introduce Trap QA mechanisms to stress-test clinical reliability by penalizing hallucinations and cognitive sycophancy. Experiments reveal current MLLMs struggle with egocentric intent due to over-reliance on global features, leading to fabricated observations and uncritical acceptance of invalid instructions.
