This research introduces an innovative approach that leverages large language models to improve medication recommendation systems in healthcare settings.

• Addresses the cold-start problem for first-time hospital patients
• Captures nuanced semantic information in medical data beyond simple identifiers
• Demonstrates superior performance on major healthcare datasets (MIMIC-III and MIMIC-IV)
• Provides more personalized treatment recommendations aligned with patient-specific conditions

This advancement matters because more accurate medication recommendations directly impact patient outcomes, reduce adverse drug events, and improve healthcare efficiency—particularly for patients without established medical histories.

Large Language Model Distilling Medication Recommendation Model