Abstract
Abstract
Purpose
Intraoperative dialogue between faculty and trainees is rich with feedback; however surgical training still relies primarily on delayed, subjective post-procedural written evaluations. We sought to capture and analyze minute-to-minute intraoperative communication to evaluate how trainee proficiency alters dyadic interaction. This pilot study aimed to demonstrate the feasibility of operative linguistic capture and analysis and map specific operative linguistic patterns to attending-rated trainee entrustability.
Methods
Audio recordings of attending-trainee interactions were captured during 25 general surgery operations. Dialogue was transcribed and 38 unique linguistic codes were developed to categorize all utterances. Post-operative resident autonomy was rated by attendings using an EPA-based scale. We compared linguistic patterns in lower-autonomy (direct supervision) versus higher-autonomy (indirect supervision or practice-ready) cases using automated tools verified by investigators.
Results
More autonomous learners contributed more intraoperative speech than less autonomous learners (28.4% v 10.8%,
p
p
p
p
0.04). Total feedback events per case were similar (
p
= 0.4), but attendings in higher-autonomy cases delivered more positive feedback (63.8% v 24.0%,
p
p
Conclusions
Distinct, quantifiable speech signatures—characterized by resident-led control of case flow, shared mental modeling, and less step-by-step technical instruction—mark higher EPA autonomy ratings. Capturing OR dialogue provides a behavior-based supplement to EPA ratings, offering a scalable method to support targeted faculty coaching and objective assessments of practice readiness.