Vocal Loudness and Stability as Behavioral Indicators of Surgeons’ Non-Technical Skills

Abstract

Non-technical skills (NTS) are essential for safe surgical teamwork, while objective measures of how surgeons communicate remain limited. This pilot study examined whether sentence-level vocal loudness and stability distinguish simulated good- and bad-NTS conditions and identified the voice features most sensitive to NTS differences. Three surgical education fellows completed three scenarios under both conditions. Good- and bad-NTS events were designed to reflect NOTSS scores of 4 and 1. Surgeon speech was isolated, vocal intensity was quantified using root-mean-square (RMS) amplitude, and sentences were classified as low, normal, or high volume relative to event-specific reference levels. Vocal stability was assessed using state entropy and state-persistence probability. Bad-NTS events showed greater loudness, more frequent extreme-volume states, higher entropy, and lower persistence, with moderate-to-large effects (|δ| ≥ .60). Mean event-level RMS, high-volume-state frequency, state entropy, and state persistence were the most sensitive features (|δ| > .80), supporting their use in objective NTS assessment.