Entrustability Meets Surgical Process Modeling: Case-Specific Prediction of Operative Autonomy

INTRODUCTION: Surgical education is evolving towards a competency-based paradigm, relying on measuring resident entrustability. Entrusting a trainee to perform a procedure independently can be a difficult judgment task, requiring predicting the various paths how an operation can play out in the OR, and estimating a trainee’s current skill level across those potential paths. This task can benefit from a decision aid, incorporating surgical process modeling to map the complexity of performing procedures, and learning curve modeling to quantify operative autonomy across trainees. METHODS: Teaching faculty completed autonomy-based operative performance evaluations via a HIPAA-compliant, web-based education management platform synchronized with operative schedules, across three general surgery residency programs from February 2017 to February 2022. These evaluations were used to construct Bayesian learning curves for each resident for various procedures, including appendectomy, cholecystectomy, colectomy, and inguinal hernia repair. Procedure approach and operative technique variants were modeled using surgical process modeling based on Markov decision processes, adaptive to the case type, surgical team, and likelihood of conversion. RESULTS: Two thousand twenty-five operative learning curves were created from 6,064 operative performance evaluations by 111 faculty members for 221 general surgery residents. These learning curves enabled an entrustability decision aid applied on 19,159 resident cases during the study period (Figure).FigureCONCLUSION: Competency-based standards in surgical education rely on entrustable professional activities (EPAs). Characterizing resident entrustability for the operative phase of an EPA is essential but difficult to determine preoperatively with confidence. Decision models allow for personalization and optimization of each resident’s pathway towards surgical competency.