Progressive decline in EEG low frequency spectral power during overnight sleep

Electroencephalography (EEG) is a cornerstone in the analysis of sleep architecture and sleep disorders. Prior studies describe decreases in EEG power towards the end of overnight sleep. However, most of these observations are based on small samples, rely in part on qualitative analyses, do not characterize EEG power at intermediate stages, and do not examine all sleep stages. Importantly, the possible relation between a dynamic change in EEG power and health outcomes was not explored. We sought to compute power in sleep stages for the first, middle, and last instances of sleep using automated, quantitative EEG methods leveraging 10,000 high-resolution overnight polysomnography cases. Results confirm that power in the 1–20 Hz range during the last instances of all sleep stages is significantly decreased compared to first instances. However, power increases in the delta and theta bands during the middle instances of N1 and N2. For N1, the decline in power follows a linear trend. These changes were not uniform across brain regions, and correlations between the EEG power decrease and clinical outcomes were statistically significant albeit with modest effect size. Thus, our work provides a rigorous, large-scale analysis of temporal trends of EEG power across the overnight sleep period, revealing novel dynamics and a potential impact on health beyond sleep disorders.