MERA leverages large language models to improve early disease detection from patient medical histories, addressing data scarcity and complex disease spaces.

• Combines memorization techniques with sophisticated ranking mechanisms
• Enhances clinical diagnosis prediction despite limited patient data
• Outperforms traditional approaches on major clinical datasets (MIMIC-III and IV)
• Creates more reliable, explainable diagnostic recommendations

This research enables more accurate early disease detection, potentially improving patient outcomes through timely intervention and treatment planning in healthcare settings.

Original Paper: Memorize and Rank: Elevating Large Language Models for Clinical Diagnosis Prediction