Multimodal physiological correlates of surgeon stress in live robot-assisted surgery

Robot-assisted surgery (RAS) imposes cognitive and psychomotor demands on console surgeons, but the physiological correlates of perceived stress during live procedures remain insufficiently characterized. Prior work has often relied on simulations or broad phase-based assessments, providing limited insight into short-term stress fluctuations in the operating room.

We conducted a prospective observational field study during live RAS procedures. Seven senior urologists were monitored during one procedure each using wireless electroencephalography (EEG), surface electromyography (sEMG), and electrocardiography (ECG). ECG-derived heart rate variability (HRV) was calculated. To avoid disrupting surgical workflow, surgeons retrospectively annotated perceived stress using video-stimulated recall (VSR) of console-view videos. Associations between physiological features and stress ratings were assessed using a linear mixed-effects model.

Across 151 VSR-based annotation windows, higher perceived stress was associated with a higher beta-to-alpha power ratio at the central midline EEG channel (Cz; β = 0.42, 95% confidence interval [CI] 0.08 to 0.75), greater upper trapezius sEMG activity (β = 0.37, 95% CI 0.06 to 0.68), shorter mean normal-to-normal cardiac intervals (Mean NN; β = - 0.39, 95% CI - 0.70 to - 0.07), and lower standard deviation of normal-to-normal intervals (SDNN; β = - 0.44, 95% CI - 0.86 to - 0.01). These directions were generally consistent across robustness and sensitivity analyses. Other EEG spectral and HRV features were not significantly associated with perceived stress.

In live RAS, perceived stress reported by console surgeons was associated with selected EEG, sEMG, and HRV features. These findings suggest that combining VSR-based stress annotations with limited physiological monitoring is feasible during live RAS and may help identify high-demand operative moments during postoperative review.