Analyzing instructor facilitation patterns in debriefings using AI-based speech metrics

Abstract

Background Facilitation of debriefings is critical for fostering reflection and behavioral change. However, little is known about how novice instructors distribute their speaking time and how this impacts perceived debriefing quality. Artificial intelligence (AI)-driven conversational analysis offers an objective method to characterize facilitation patterns. Methods Participants were fifteen novice healthcare instructors with less than one year of experience who recently completed an online simulation course. A cross-sectional study was conducted using GailBot API for automated transcription and speech analysis of remote interprofessional debriefings. Standardized scenarios and consistent student participants controlled external variables. Instructor facilitation patterns were compared to DASH evaluations from students, instructors, and an expert observer. Results Instructors spoke for an average of 56.09% of debriefing time. Higher instructor talk time correlated negatively with DASH self-ratings (Spearman’s Rho = −0.627, p = .012). Conclusion AI-based analysis provides valuable insights into facilitation patterns, supporting multidisciplinary faculty development.