The impact of sleep deprivation on dynamic functional connectivity of the brain: Based on alertness task performance

Sleep deprivation (SD) impairs mood and cognition, yet its dynamic neural mechanisms remain unclear. Forty healthy adults (30-h SD) completed mood assessments, resting-state fMRI (rs-fMRI), and psychomotor vigilance task (PVT) tests at baseline and post-SD. Using dynamic functional connectivity (dFC) with sliding windows and k-means clustering, we identified two recurrent whole-brain states: (i) an economical state with sparse, weaker global coupling and (ii) a maladaptive compensatory state with globally strengthened synchronization. SD increased both the fraction of windows and mean dwell time (MDT) of the maladaptive state. Across participants, PVT lapses correlated positively with the maladaptive state's MDT and fraction of windows and negatively with those of the economical state. Finally, we built an interpretable predictive model of PVT lapses using competitive adaptive reweighted sampling partial least-squares regression (CARS-PLSR), which highlighted connections within the dorsal attention network (DAN) as key predictors. These findings link behavioral impairment to altered brain-state dynamics and provide a sparse, testable feature set that can support early risk stratification and intervention for SD-related cognitive decline.