Decoding Wakefulness Levels from Typical fMRI Resting-State Data Reveals Reliable Drifts between Wakefulness and Sleep

The mining of huge databases of resting-state brain activity recordings represents state of the art in the assessment of endogenous neuronal activity-and may be a promising tool in the search for functional biomarkers. However, the resting state is an uncontrolled condition and its heterogeneity is neither sufficiently understood nor accounted for. We test the hypothesis that subjects exhibit unstable wakefulness, i.e., drift into sleep during typical resting-state experiments. Analyzing 1,147 resting-state functional magnetic resonance data sets, we revealed a reliable loss of wakefulness in a third of subjects within 3 min and demonstrated the dynamic nature of the resting state, with fundamental changes in the associated functional neuroanatomy. Implications include the necessity of wakefulness monitoring and modeling, taking measures to maintain a state of wakefulness, acknowledging the possibility of sleep and exploring its consequences, and especially the critical assessment of possible false-positive or false-negative results.