LLM-Match introduces a novel framework that uses fine-tuned large language models to accurately match patients with appropriate clinical trials based on their medical records.

Key Innovations:

• Retrieval-Augmented Generation (RAG) extracts relevant patient context from electronic health records
• Specialized prompt generator creates tailored inputs for the matching process
• Fine-tuned open-source LLMs deliver accurate matching capabilities with lower costs
• Comprehensive evaluation across multiple medical datasets demonstrates effectiveness

Why It Matters: This research addresses critical healthcare challenges by improving the efficiency and accuracy of matching patients to clinical trials, potentially accelerating medical research while ensuring appropriate patient care. The approach demonstrates how AI can transform healthcare operations with practical, cost-effective solutions.

LLM-Match: An Open-Sourced Patient Matching Model Based on Large Language Models and Retrieval-Augmented Generation