The quality of sleep and its cognitive benefits relies on the cyclic alternance of Non-rapid-eye-movement (NREM) and REM sleep. The ability to predict shifts in sleep stages could help design future interventions in sleep medicine, but it remains unknown how robust the NREM-REM sleep architecture may be for a given individual over many nights. We sought to define the individual variability and predictability of healthy human sleep recorded over unprecedented durations (weeks).
Based on ultra-long-term subscalp electroencephalographic (sqEEG) recordings from a newly-available, minimally invasive device, we characterized sleep cycles in eight healthy subjects over a median of 30 consecutive days. We first decomposed EEG signals into five frequency bands (δ, θ, α, σ, and β) using a multitaper time-frequency transform. Second, we quantified variability in sleep spectral composition and predictability in sleep stage transitions using unsupervised (dynamic time warping) and supervised (generalized linear models) learning methods, respectively.
Using dynamic time warping, we quantified the dissimilarity (D) between pairs of nights, showing that it was higher within (D = 2.5 ± 0.7) than across subjects (D = 4.1 ± 0.5, p < 0.001). Furhter, we extracted archetypal sleep patterns, which are most representative of the average NREM-REM spectral architecture. Based on the found interplay between δ and σ power bands, we predicted transitions from NREM to REM two minutes in advance with high accuracy (area under the receiver operating characteristic curve equal to 0. 88).
Taken together, these results show that sleep is variable night after night in healthy subjects, but that the dynamics in core sleep oscillations are shared across individuals. As a translational outlook, the predictability of certain sleep transitions affords the means to anticipate pathological symptoms specific of a given sleep stage.
