Resting-state aperiodic neural activity as a novel objective marker of daytime somnolence

Study ObjectivesCurrent assessment of excessive daytime somnolence (EDS) requires subjective measurements such as the Epworth Sleepiness Scale (ESS), and/or resource intensive sleep laboratory investigations. Recent work 1,2 has called for more non-performance-based measures of EDS. One promising non-performance-based measure of EDS is the aperiodic component of electroencephalography (EEG). Aperiodic (non-oscillatory) activity reflects excitation/inhibition ratios of neural populations and is altered in various states of consciousness, and thus may be a potential biomarker of hypersomnolence. MethodsWe retrospectively analysed EEG data from patients who underwent a Multiple Sleep Latency Test (MSLT) and determined whether aperiodic neural activity is predictive of EDS. Participants having undergone laboratory polysomnogram and next day MSLT were grouped into MSLT+ (n = 26) and MSLT- (n = 33) groups (mean sleep latency of 10min, respectively) and compared against a non-clinical (Control) group of participants (n = 26). ResultsWhile the MSLT+ and MSLT- groups did not differ in their aperiodic activity, the Control group had a significantly flatter slope and larger offset compared to both MSLT+ and MSLT- groups. Logistic regression machine learning predicted group status (i.e., symptomatic, non-symptomatic) with 90% accuracy based on the aperiodic slope while controlling for age. Slow oscillation-spindle coupling was also significantly stronger in the Control group relative to MSLT+ and MSLT- groups. ConclusionsOur results provide first evidence that aperiodic neural dynamics and sleep-based cross-frequency coupling is predictive of EDS, thereby providing a novel avenue for basic and applied research in the study of sleepiness.