Decoding surgical proficiency and complexity: a machine learning framework for robotic herniorrhaphy.

Area of Science:

• Robotic surgery
• Surgical education
• Machine learning in medicine

Background:

• Assessing surgical case complexity is crucial for operative planning and patient outcomes.
• Current complexity metrics often overlook operator performance, relying on external clinical factors.
• Robotic surgery offers novel, quantifiable performance data for surgical skill assessment.

Purpose of the Study:

• To determine if objective performance indicators (OPIs) can predict case complexity in robotic ventral herniorrhaphy.
• To explore how OPIs quantify surgical skill acquisition during this procedure.
• To address the gap in understanding the relationship between OPIs, case complexity, and skill development.

Main Methods:

• Analysis of OPI and clinical data from 561 robotic ventral hernia repairs.
• Application of iterative ensemble machine learning models to predict case complexity.
• Utilizing dimensional reduction and Euclidean distances to track skill evolution.

Main Results:

• Machine learning models integrating clinical and OPI data achieved an F1 score of 0.87 for complexity prediction.
• OPIs alone had an F1 score of 0.58 for complexity prediction.
• Longitudinal OPI analysis showed stabilization within 10 months, indicating skill acquisition compensated for increasing complexity.

Conclusions:

• OPIs alone are weak predictors of surgical case complexity.
• Temporal OPI evolution effectively demonstrates surgical skill acquisition.
• Digital metrics from robotic surgery provide insights into the interplay between surgeon learning and case difficulty.