GLoC-EHR：基于全局上下文与局部EHR事件的证据引证临床推理
GLoC-EHR: Evidence-Cited Clinical Reasoning over Global Context and Local EHR Events
- Seoul National University College of Medicine（首尔大学医学院）
- Seoul National University Hospital（首尔大学医院）
- Healthcare AI Research Institute, Seoul National University Hospital（首尔大学医院医疗AI研究所）
机构由 AI 辅助整理，请以论文原文为准。
中文总结 AI 辅助
GLoC-EHR是一种多模态语言模型，通过全局和局部记忆编码EHR轨迹，经GRPO训练实现证据引证推理，在MIMIC-IV任务上取得最高宏AUROC，并成功迁移至EHRSHOT。
中文摘要 AI 辅助
结构化电子健康记录（EHR）包含患者以临床代码序列形式呈现的临床轨迹。从这些记录回答临床问题需要整个轨迹的上下文以及支持答案的特定事件。我们引入了GLoC-EHR，一种多模态语言模型，它通过固定大小的全局记忆（用于轨迹）和局部记忆（用于选定事件）读取记录的上下文编码。该模型学习从全局记忆生成住院病程摘要，并从局部记忆生成掩蔽概念的描述，将两者与临床文本对齐。随后，它被训练在回答之前引证证据，通过基本原理微调，随后进行组相对策略优化（GRPO），奖励正确回答和记录支持的证据。在三个MIMIC-IV结果任务中，当直接回答时，GLoC-EHR在比较模型中取得了最高的宏AUROC，而读取序列化记录的零样本LLM则远远落后。通过证据引证推理，它在宏AUROC上接近其直接多任务对应模型，并且目标的证据项在相似的宏AUROC下减少了无支持的证据。局部记忆增加了独特的支持发现，特别是在严格匹配下，而没有可检测的宏AUROC变化。无需重新训练，GLoC-EHR在EHRSHOT上的表现与EHR-BERT相当，并且比其自身骨干的零样本提示更好地回答了两个未见过的实验室问题。
英文摘要
Structured electronic health records (EHRs) contain a patient's clinical trajectory as a sequence of clinical codes. Answering clinical questions from such records requires both the context of the whole trajectory and the specific events that support the answer. We introduce GLoC-EHR, a multimodal language model that reads a contextual encoding of the record through a fixed-size global memory of the trajectory and a local memory of selected events. The model learns to generate hospital-course summaries from the global memory and descriptions of masked concepts from the local memory, aligning both with clinical text. It is then trained to cite evidence before answering, through rationale fine-tuning followed by group relative policy optimization (GRPO) with rewards for correct answers and record-supported evidence. On three MIMIC-IV outcome tasks, GLoC-EHR attains the highest macro AUROC among the compared models when it answers directly, whereas zero-shot LLMs reading the serialized record fall far behind. With evidence-cited reasoning, it stays close to its direct multi-task counterpart in macro AUROC, and the evidence terms of the objective reduce unsupported evidence at a similar macro AUROC. The local memory adds distinct supported findings, particularly under strict matching, without a detectable change in macro AUROC. Without retraining, GLoC-EHR transfers to EHRSHOT on par with EHR-BERT and answers two unseen laboratory questions better than zero-shot prompting of its own backbone.
