medIKAL is a novel framework that integrates Large Language Models with knowledge graphs to enhance diagnostic capabilities when analyzing complex Electronic Medical Records (EMRs).

• Assigns weighted importance to medical entities based on type, enabling more precise analysis
• Performs local contextualization to focus on relevant information within lengthy medical records
• Leverages knowledge graphs to provide structured medical knowledge that complements LLMs' capabilities
• Demonstrates significant improvements in diagnostic accuracy compared to standard LLM approaches

This research addresses critical challenges in healthcare information processing, potentially reducing diagnostic errors and improving patient outcomes through more accurate interpretation of complex medical data.

Original Paper: medIKAL: Integrating Knowledge Graphs as Assistants of LLMs for Enhanced Clinical Diagnosis on EMRs