EMERGE introduces a novel framework that combines multimodal Electronic Health Records with retrieval-augmented generation to significantly improve clinical predictive accuracy.

• Seamlessly integrates clinical notes and time-series data with contextual medical knowledge
• Employs retrieval-augmented generation (RAG) to enhance model understanding of patient conditions
• Demonstrates superior performance on in-hospital mortality and readmission prediction tasks
• Provides more explainable and transparent clinical predictions

This research matters because it addresses critical healthcare challenges by improving prediction accuracy while maintaining interpretability, potentially reducing medical errors and improving patient outcomes in clinical settings.

EMERGE: Enhancing Multimodal Electronic Health Records Predictive Modeling with Retrieval-Augmented Generation