Portable polysomnography-derived multi-domain EEG markers of anesthetic state transitions during propofol anesthesia for gastrointestinal endoscopy.

Abstract

Background: Propofol anesthesia is widely used for gastrointestinal (GI) endoscopy, but electroencephalographic (EEG) features of anesthetic state transitions in this brief and stimulus-rich setting remain incompletely characterized. This study aimed to describe clinically interpretable EEG markers across anesthetic stages using portable polysomnography (PSG)-derived six-channel EEG.

Methods: In this prospective observational study, 31 adults undergoing GI endoscopy under propofol-based anesthesia underwent continuous portable PSG recording. EEG analyses focused on six scalp channels (F3, F4, C3, C4, O1, and O2) across four clinically defined stages: baseline, induction, maintenance, and recovery. Relative spectral power, spectral edge frequency at 95% cumulative power (SEF95), and spindle-like event metrics were quantified. Sensor-level functional connectivity was assessed using weighted phase-lag index, imaginary coherence, and amplitude envelope correlation (AEC). Spectral and spindle-like event measures were analyzed using linear mixed-effects models, and connectivity differences were assessed using permutation-based paired tests with false discovery rate correction.

Results: Relative delta power decreased from baseline and induction to maintenance and recovery, whereas theta, alpha, sigma, and beta relative power and SEF95 generally increased. Spindle-like event density differed across stages and was highest during recovery, while mean duration and peak frequency remained relatively stable within the 11-16 Hz range. Functional connectivity showed metric-specific modulation: AEC demonstrated the most widespread stage-related differences, whereas phase-based connectivity measures showed only limited edge-specific changes after correction for multiple comparisons.

Conclusions: Portable PSG-derived six-channel EEG captured stage-related spectral, spindle-like event, and sensor-level connectivity changes during propofol anesthesia for GI endoscopy. These findings identify candidate EEG features for future evaluation of emergence detection and atypical anesthetic-state identification during procedural anesthesia. Their utility for real-time propofol titration or excessive-sedation warning requires prospective validation.