Summary:
UCLA researchers in the Department of Surgery have developed a deep-learning architecture to efficiently and accurately assess surgeon skill level for improved medical training.
TITLE: AI-Powered System for Objective Surgical Skill Assessment in Open Procedures
Background:
Objectively assessing surgical trainee performance during open procedures remains a significant challenge in medical education. Traditional evaluations rely heavily on subjective expert observation, introducing variability and bias that undermine standardized training and longitudinal skill tracking. These inconsistencies directly impact training feedback quality and create barriers to cross-institutional standardization of surgical education.
While minimally invasive procedures benefit from controlled environments with standardized camera angles, open surgery presents unique assessment challenges due to unrestricted movement and variable viewing perspectives. Additionally, existing methods struggle to capture subtle skill transitions, particularly at intermediate levels. Healthcare systems worldwide require more rigorous, data-driven approaches to uphold surgical competency standards and ensure patient safety. An innovative solution must enhance assessment accuracy, consistency, and scalability while seamlessly integrating into current training programs.
Innovation:
Researchers at UCLA’s Department of Surgery have developed a novel two-phase deep learning architecture to automate surgical skill assessment, providing immediate, objective feedback without requiring constant expert observation. The system uses a two-phase deep learning approach: first, it analyzes surgical techniques through a specialized computer vision model trained to recognize subtle movements and technical precision. Then, an advanced AI model evaluates the overall procedure, identifying key moments, such as suture placements and tissue handling, by assigning importance to different steps, similar to how expert surgeons assess performance, while maintaining awareness of overall procedural flow. This system adapts to different surgical perspectives and focusing distances, overcoming a significant limitation in open surgery assessment. Unlike traditional methods that struggle with nuanced skill transitions, particularly at intermediate proficiency levels, this model provides a hierarchical and progressive understanding of skill development. It maintains logical consistency in its assessments, accurately distinguishing between novice and expert performances while offering nuanced differentiation within skill levels. Additionally, this invention delivers immediate, objective feedback without requiring specialized tracking devices or sensor-instrumented tools, ensuring a cost-effective, scalable solution that seamlessly integrates into existing training programs. By reducing reliance on expert observation time, providing interpretable results, and enabling targeted feedback, this system represents a transformative advancement in surgical education, making high-quality, evidence-based training accessible to a broader range of institutions.
Potential Applications:
• Surgical education and training
• Competency-based certification
• Hospital Quality Assurance
• Remote and Tele-Surgical Training
• Healthcare Policy and Regulation Compliance
Advantages:
• Objective and consistent evaluation, eliminating evaluator bias and ensuring standardized skill assessments
• Automated, immediate, data-driven feedback
• Enhanced surgical training
• Adaptability to open surgery, handling unrestricted movement and varying camera angles
• Supportive of longitudinal training and development, aiding in curriculum development and evaluations
• Scalability and easy implementation, reducing integration barriers
State of Development:
This invention was publicly disclosed in a conference presentation on February 11, 2025.
References:
UCLA Case No. 2025-207
Lead Inventor:
Peyman Benharash, UCLA Professor of Surgery.