arXiv：cs.LG（机器学习，全量分类）· Chaiho Shin, Kwangsoo Kim·· 15 小时前AI 评分34
GLoC-EHR：基于全局上下文与本地 EHR 事件的循证临床推理模型
GLoC-EHR: Evidence-Cited Clinical Reasoning over Global Context and Local EHR Events
AI 导读
GLoC-EHR 是一款多模态语言模型，通过固定大小的全局记忆与本地事件记忆读取结构化 EHR，在三个 MIMIC-IV 结局任务上直接作答时取得对比模型中最高的 macro AUROC，零样本 LLM 读取序列化记录则远落后。
正文
Abstract: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.
| Subjects: | Machine Learning (cs.LG) |
| Cite as: | arXiv:2610.01076 [cs.LG] |
| (or arXiv:2610.01076v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.01076 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Chaiho Shin [view email]
[v1]
Thu, 1 Oct 2026 05:16:20 UTC (283 KB)
来源：arXiv：cs.LG（机器学习，全量分类） · arxiv.org