MicroRAG: Development of a Novel Artificial Intelligence Retrieval-Augmented Generation Model for Microsurgery...

Area of Science:

• Microsurgery and Artificial Intelligence
• Clinical Decision Support Systems
• Medical Literature Synthesis

Background:

• Microsurgical decisions integrate patient factors, techniques, and intraoperative data.
• Existing AI, large language models (LLMs), and retrieval-augmented generation (RAG) lack microsurgery-specific clinical decision support.
• MicroRAG is the first AI system for microsurgery, providing instant, evidence-based recommendations from the literature.

Purpose of the Study:

• To develop and evaluate MicroRAG, an AI-powered clinical decision support system for microsurgery.
• To leverage retrieval-augmented generation (RAG) technology for real-time, literature-backed guidance.
• To address the need for integrated decision support in complex microsurgical scenarios.

Main Methods:

• Developed an AI system integrating 4876 microsurgical publications (2000-2024) using RAG.
• Implemented hierarchical document clustering for query processing.
• Evaluated performance on 10 clinical scenarios measuring relevancy, faithfulness, and accuracy.

Main Results:

• MicroRAG achieved high relevancy (0.953) and faithfulness (0.907) scores.
• System demonstrated strong correctness (0.88) and semantic similarity (0.75).
• Provided actionable guidance for free flap monitoring, complication management, and technique selection, all literature-grounded.

Conclusions:

• MicroRAG offers a technological innovation for microsurgery, enabling instant access to evidence-based recommendations.
• The system can standardize best practices and reduce decision-making uncertainty.
• Potential to improve patient outcomes across all experience levels in microsurgery.