From Brain Networks to Sleep Perception: EEG-Derived Global Efficiency is Associated with Subjective Sleep Quality in University Students.

Abstract

Objective: The correspondence between objective sleep physiology and subjective sleep experience remains poorly understood. Conventional polysomnography (PSG) primarily captures sleep macrostructure and may fail to reflect large-scale neural integration processes underlying perceived sleep quality. This study examined whether sleep-stage-specific brain network integration, quantified via EEG-derived global efficiency (GE), is associated with daytime sleepiness and insomnia severity in healthy university students.

Methods: In this cross-sectional observational study, overnight full-night PSG was recorded in 52 healthy male university students. The debiased weighted Phase Lag Index was used to measure functional connectivity to minimize volume conduction effects. GE was computed across five canonical frequency bands during N3 and REM sleep. Subjective sleep outcomes were assessed using the Epworth Sleepiness Scale (ESS) and the Insomnia Severity Index (ISI). Associations between GE, subjective sleep measures, and conventional PSG parameters were evaluated using linear and multiple regression analyses.

Results: Distinct stage-dependent differences in network topology were observed, with lower alpha-band GE during REM sleep and lower delta-band GE during N3 sleep. Higher daytime sleepiness (ESS) was associated with increased GE during REM sleep in the delta, beta, and gamma bands, whereas higher theta-band GE during N3 sleep was associated with lower insomnia severity (ISI). GE-based models accounted for a modest proportion of variance in subjective outcomes (ISI: R² =0.082; ESS: R² =0.190). Conventional PSG macrostructural indices were not significantly associated with subjective sleep measures.

Conclusion: EEG-derived GE showed stage- and frequency-specific associations with subjective sleep outcomes, with limited additional contribution beyond conventional macrostructural PSG measures. These findings provide preliminary evidence that sleep network topology may partially capture aspects of subjective sleep experience not fully reflected by standard sleep architecture indices. EEG-based GE may serve as a potential network-level correlate of subjective sleep experience, suggesting a possible network-level link between sleep physiology and perceived sleep quality in non-clinical populations.