Patients-like-me: A Variational LM--GNN Framework for Explainable Clinical Prediction

2026-08-04

AI Summary

Existing language models for electronic health records often model patient sequences in isolation and lack interpretability, while graph neural networks struggle to effectively capture textual semantics. This work proposes the first unified variational framework integrating language models and graph neural networks (LM-GNN), which alternately optimizes local patient semantics and global cohort representations under a supervised objective via a variational expectation-maximization algorithm. The approach enables interpretable, similarity-based patient tracing by jointly leveraging reference patient retrieval and edge mask analysis. Evaluated on MIMIC-III and MIMIC-IV, the method significantly outperforms current state-of-the-art approaches, is compatible with diverse language model architectures, incurs low computational overhead, and accurately identifies the most influential reference patients contributing to predictions.

Abstract

Language models (LMs) offer strong textual representations for electronic health records (EHRs), but they encode patient sequences in isolation and provide limited explainability. Graph neural networks (GNNs) complement LMs by incorporating inter-patient relationships and enabling reference-patient attribution, yet they rely on high-quality patient representations. We propose Patients-like-me (PLM), a unified LM--GNN framework that integrates local patient semantics with global cohort structure. To train PLM efficiently, we introduce a Variational Expectation-Maximization algorithm that alternates LM and GNN updates under a supervised variational objective. Extensive experiments on MIMIC-III and MIMIC-IV show that PLM consistently outperforms state-of-the-art methods, with improvements generalizing across encoder-only and decoder-only LM backbones. These gains are achieved with only modest additional computational overhead. PLM also provides reference-patient explanations by retrieving influential similar patients, while edge-masking experiments confirm that the highest-ranked references have the greatest impact on model predictions.

Problem

• clinical prediction
• explainability
• language models
• graph neural networks
• electronic health records

Innovation

• Variational EM
• LM-GNN integration
• explainable clinical prediction
• reference-patient attribution
• electronic health records