Sleepiness phenomics: Modeling individual differences in subjective sleepiness profiles

Abstract

The present study investigates individual differences in subjective sleepiness profiles during 36h of sustained wakefulness in a modified constant routine protocol. Twenty-three volunteers (11 females), aged between 18 and 47yrs (M age=30.41, SD=10.26) enrolled in the study. Subjective sleepiness ratings were collected every 2h by means of visual analogue scales. Circadian rhythmicity was assessed by means of salivary cortisol. Subjective sleepiness data were analyzed using functional principal component analysis (fPCA). Our results show that approximately 80% of the variance is accounted for by three functional components. The first component explains 50.28% of the variance and is characterized by a profile of exclusively positive loadings, representing vertical shifts from the mean sleepiness profile. Scores on this component are positively related to self-reported habitual sleep times and mean slow wave activity (SWA) during wake. Positive scores on the second component (18.40% of the variance) are characterized by a higher than average peak-to-trough amplitude in subjective sleepiness profiles. Participants with higher than average scores on this component show a significantly higher amplitude in salivary cortisol profiles as opposed to participants with lower than average scores. Participants with positive scores on the third component (10.09% of the variance) show higher than average levels of subjective sleepiness during morning hours, a buildup of wake effort occurring later and more afternoon sleepiness after sleep deprivation than negative scorers. Peak levels of salivary cortisol occur significantly later in these participants. Taken together, our results suggest that component 1 represents tonic differences in sleepiness profiles primarily related to mechanisms of sleep homeostasis, component 2 to circadian amplitude differences and component 3 to diurnal preference. However, since the components are additions to a mean profile, each of the three components is likely to correspond to a mixture of multiple physiological parameters, rather than to a single process. The approach shows interesting potential for (1) revealing unidentified physiological processes, (2) testing existing assumptions about regulatory mechanisms at the basis of interindividual variability in sleepiness profiles and (3) the specification of sleepiness phenotypes on a quantitative basis.