保留 UID 的多模态临床时间线重建与 GAVEL 源证据裁定方法
Anchoring Clinical Events in Time: UID-Preserving Multimodal Reconstruction and Source-Grounded Adjudication
AI 导读
一项发表于 ML4H 2026 的工作提出临床时间线重建管线，通过检索结构化 EHR 证据为叙事事件推断时间，并给每个事件保留持久 UID 以区分重复提及；同时引入 LLM 裁判 GAVEL，对照病历与结构化记录核查两条时间线的分歧并给出证据。
正文 · 原文
Abstract:Interpreting a patient's hospital course depends on when symptoms developed, tests were performed, and treatments began. Discharge summaries often leave this chronology unclear, while structured electronic health record (EHR) tables timestamp only some of these events. We reconstruct clinical timelines by retrieving structured evidence for narrative events and jointly revising their estimated times. Each extracted occurrence carries a persistent unique identifier (UID) linked to its source span, keeping repeated mentions distinct throughout reconstruction. We also introduce GAVEL, a large language model (LLM) judge that checks disagreements between two timelines against the note and structured record and cites evidence for its verdicts. We evaluated six open-weight models on 40 clinician-annotated discharge summaries. Our pipeline recovered more reference events than LLM-Timeline and TKW2 when each comparison used the same backbone. For GLM-5.2, adding structured evidence improved timestamp agreement with essentially unchanged event recovery; ablations showed that links to timestamped source rows contributed to this improvement. Results varied across the other models, with some timing gains accompanied by fewer recovered events. Both GAVEL judges, GLM-5.2 and MiMo-V2.6-Pro-RL, favored multimodal over text-only GLM-5.2 timelines, while neither clearly distinguished the DeepSeek V3.2 variants. Manual review upheld most sampled GAVEL findings. The source links allow reconstructed times to be checked, and GAVEL identifies disagreements for human review.
| Comments: | Accepted as Proceedings of Machine Learning for Health (ML4H 2026) in PMLR |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2609.13062 [cs.AI] |
| (or arXiv:2609.13062v2 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2609.13062 arXiv-issued DOI via DataCite |
Submission history
From: Sayantan Kumar [view email]
[v1]
Fri, 11 Sep 2026 17:01:10 UTC (10,331 KB)
[v2]
Wed, 11 Nov 2026 17:19:45 UTC (2,110 KB)
来源：arXiv cs.AI · arxiv.org