用双曲几何建模病历数据，小模型实现大模型问答效果
HypEHR: Hyperbolic Modeling of Electronic Health Records for Efficient Question Answering
- 将诊断码、就诊记录和问题嵌入双曲空间，利用几何结构增强表示
- 在两个MIMIC-IV基准上接近大模型表现，参数量显著更少
- 适合资源有限但需高效医疗问答的场景
电子健康记录（EHR）问答通常依赖成本高昂的大型语言模型（LLM）流水线，且未显式利用临床数据的层次结构。鉴于医学本体和患者轨迹具有双曲几何特征，我们提出HypEHR，一种紧凑的Lorentzian模型，将编码、就诊记录和问题嵌入双曲空间，并通过类型特定指针头的几何一致交叉注意力回答问题。HypEHR通过下一就诊诊断预测预训练和层次感知正则化，使表示与ICD本体对齐。在基于MIMIC-IV的两个EHR问答基准上，HypEHR表现接近基于LLM的方法，但使用远少的参数。代码已公开于https://github.com/yuyuliu11037/HypEHR。
原文摘要 · Abstract (English)
Electronic health record (EHR) question answering is often handled by LLM-based pipelines that are costly to deploy and do not explicitly leverage the hierarchical structure of clinical data. Motivated by evidence that medical ontologies and patient trajectories exhibit hyperbolic geometry, we propose HypEHR, a compact Lorentzian model that embeds codes, visits, and questions in hyperbolic space and answers queries via geometry-consistent cross-attention with type-specific pointer heads. HypEHR is pretrained with next-visit diagnosis prediction and hierarchy-aware regularization to align representations with the ICD ontology. On two MIMIC-IV-based EHR-QA benchmarks, HypEHR approaches LLM-based methods while using far fewer parameters. Our code is publicly available at https://github.com/yuyuliu11037/HypEHR.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。