Back to 2026 Abstracts
Development Of A Novel Artificial Intelligence Clinical Decision Support System For Microsurgery
Berk B. Ozmen, M.D.1, Nishant Singh, MSc
2, Kavach Shah, MSc
2, Ibrahim Berber, MSc
3, Damanjit Singh, MSc
2, Eugene Pinsky, PhD
2, Graham S. Schwarz, MD, MSE, FACS
1.
1Cleveland Clinic, Cleveland, OH, USA,
2Boston University, Boston, MA, USA,
3Case Western Reserve University, Cleveland, OH, USA.
PURPOSE: Microsurgical decision-making requires integration of patient-specific factors, advanced technical considerations, and real-time intraoperative insights. Despite rapid advances in artificial intelligence (AI), large language models (LLMs), and retrieval-augmented generation (RAG) frameworks, no AI-driven clinical decision support system currently exists for microsurgery. We developed MicroRAG, the first specialty-specific AI system designed to deliver real-time, evidence-based recommendations with direct literature citations.
METHODS: A total of 4,876 peer-reviewed microsurgical publications (2000-2024) were integrated into a RAG-based decision support architecture. MicroRAG processes free-text clinical queries through hierarchical document clustering and semantic retrieval to generate concise, evidence-grounded responses with linked citations. System performance was evaluated using ten standardized clinical scenarios representing common microsurgical decisions. Outputs were assessed for relevancy, faithfulness to source literature, and clinical accuracy using quantitative evaluation metrics.
RESULTS: MicroRAG achieved an average answer relevancy of 0.953 (range: 0.857-1.000) and faithfulness of 0.907 (range: 0.676-1.000). G-Eval correctness averaged 0.88, with semantic similarity and confidence scores of 0.75 and 0.80, respectively. The system consistently provided comprehensive, verifiable guidance for complex scenarios, including free flap monitoring, vascular complication management, and surgical technique selection, supported by direct citations from peer-reviewed literature.
CONCLUSION: MicroRAG represents a technological innovation in microsurgical practice by providing immediate, literature-linked clinical guidance that traditionally requires extensive manual review. This system has the potential to standardize best practices, enhance clinical decision-making, and improve clinical outcomes.
Back to 2026 Abstracts
