首页 > AI前沿 > REFINE: LLM Refinement over Budgeted Text-Attributed Graphs for Personalized Medical Concept Representation

REFINE: LLM Refinement over Budgeted Text-Attributed Graphs for Personalized Medical Concept Representation

arXiv机器学习 2026-09-04 03:29 16 阅读 查看原文

Learning rich medical concept representations is essential for EHR prediction.

Text-attributed knowledge graphs (TKGs) provide a natural foundation by organizing heterogeneous medical relations together with textual semantics.

However, most existing encoders process concepts uniformly across patients, despite the fact that a code's meaning and predictive value depend on patient-specific clinical context and trajectory.

Learning patient-personalized concept representations from TKGs introduces two key challenges:

  1. deciding how much KG context to incorporate for each observed code,
  2. aligning semantic information with the patient-specific relational structure.

We propose REFINE, a KG-aware budgeted LLM graph refinement framework for patient-personalized medical concept encoding.

Starting from a global TKG, REFINE constructs patient-specific temporal graphs.

A sequential reinforcement learning policy selects a personalized KG expansion budget for each observed code.

The resulting patient graph is processed by a heterogeneous GNN to capture relation-aware structural dependencies, while a frozen LLM uses graph-aware soft prompts to semantically refine concept representations.

Experiments on MIMIC-III and MIMIC-IV show that REFINE consistently improves diverse EHR backbones, outperforms strong baselines, and demonstrates robust gains across component ablation, KG selection, and data insufficiency.