This research introduces a novel approach to surgical workflow analysis that mimics how humans learn: by watching experts and understanding their explanations.

• Leverages video-language models to address the lack of annotated surgical datasets
• Combines visual learning with expert knowledge transfer through language
• Enables more accurate and comprehensive surgical workflow analysis
• Addresses critical data sparsity and heterogeneity challenges in medical AI

This advancement matters for healthcare by enhancing surgical education, supporting clinical decision-making, and improving research capabilities—all without requiring extensive manual annotation of surgical videos.

Watch and Learn: Leveraging Expert Knowledge and Language for Surgical Video Understanding