Abstract
This study introduces Reliability Volume (RV), an integrated metric combining trajectory similarity with empirical reliability estimation using threshold counts to evaluate surgical skill during repetitive training. RV quantifies both spatial precision and the probability of consistent task execution, addressing limitations of single-session metrics that neglect fatigue and performance drift. Applied to knot-tying with assistive devices, RV jointly reflects spatial accuracy and performance consistency over multiple sessions. Our results show that RV reliably tracks learning progression and is readily compatible with real-time (closed-loop) feedback systems, providing a dynamic, comprehensive, and practice-oriented assessment framework.