This research combines medical knowledge graphs with large language models to create more accurate diagnostic prediction systems from Electronic Health Records (EHRs).

• Addresses the challenge of information overload from complex EHR narratives
• Leverages UMLS medical knowledge to enhance LLM reasoning capabilities
• Demonstrates how structured medical knowledge improves diagnostic accuracy
• Creates foundation for AI-augmented diagnostic decision support systems

This innovation matters because it could reduce diagnostic errors, improve healthcare efficiency, and provide more reliable decision support for clinicians navigating complex patient data.

Leveraging Medical Knowledge Graphs Into Large Language Models for Diagnosis Prediction: Design and Application Study