Recurrent memory compresses patient histories for efficient e health predictions
ReLMem: Learning Recurrent Memory for Longitudinal EHR Modeling
Machine LearningArtificial Intelligence
Summary
Electronic health records store a lot of patient visits, but keeping track of all this information can be slow and use a lot of computer memory. The authors created a system called ReLMem that summarizes each new visit into a small, updated memory without needing to reread older records. This method helps keep important details from past visits while using much less memory. Their results show that ReLMem can predict patient medication well, almost as accurately as methods that use the full history, but with far less data stored.
What this means in practice
- •For hospital data teams: Compress patient histories to enable fast and memory-efficient medication predictions on large electronic health record systems.
- •For healthcare software developers: Integrate recurrent memory modules into clinical AI tools to maintain updated patient state without reprocessing entire histories for each prediction.
Authors
Zijie Meng, Xiwei Dai, Yingying Zhang, Jian Wu, Xian Wu, Zuozhu Liu
Abstract
Longitudinal electronic health record (EHR) modeling requires integrating new visits with an expanding patient history. Yet the continual accumulation of clinical information imposes increasing computational and memory costs on large language models (LLMs) when they process and retain complete patient histories. A practical alternative is visit-wise recurrent compression, which incorporates each incoming visit into a compact, continually updated patient memory. However, under a fixed memory budget, successive updates must integrate new information without progressively losing critical historical evidence needed to subsequent tasks. To address this challenge, we introduce Recurrent Longitudinal Memory (ReLMem), a framework that learns to maintain fixed-capacity patient memory for efficient downstream prediction with a frozen LLM. ReLMem equips this LLM with lightweight compression adapters to recurrently update the memory from its previous state and each incoming visit, without rereading earlier records. Specifically, we develop a multi-granularity optimization strategy to preserve task-relevant information throughout recurrent updates and support downstream prediction from the final memory. The intermediate supervision aligns attention outputs from compressed memory and the full history under identical queries, while prediction supervision minimizes cross-entropy with ground truth answers conditioned on the final memory. On EHR-based medication prediction, ReLMem approaches the F1 scores of full-history baseline while reducing average retained historical storage by 97.1%. Under the same memory budget, it improves macro- and micro-F1 over the strongest compressed-memory baseline by 4.66 and 4.75 percentage points, respectively. These results highlight the value of learning recurrent patient memory for efficient longitudinal EHR modeling.