2.5. Diagnostic Polysomnography (PSG)

PSG was performed at the American Academy of Sleep Medicine (AASM)-Accredited Massachusetts General Hospital Sleep Disorders Unit. During the COVID-19 pandemic, all participants were cleared with PCR COVID-19 testing 72 h prior to the study. A minimum of six hours of overnight sleep was monitored using conventional in-lab polysomnography (Compumedics, Charlotte, NC, USA) with a 250 Hz sampling rate and a 0.3–35 Hz bandpass filter. EEG data were recorded from frontal (F3 and F4), central (C3 and C4), and occipital (O1 and O2) electrodes, and then referenced to contralateral mastoid electrodes (M2 and M1). Electrooculogram (EOG), electromyogram (EMG), ECG, pulse oximetry, respiration, and nasal flowmeter were used to assess sleep apnea and leg movements. Sleep was staged as non-rapid eye movement (NREM) stages 1 to 3, R (REM), or awake (W) stages in consecutive 30 s epochs following AASM criteria by Registered Polysomnographic Technologists [37]. The final sleep report was reviewed by a physician with board certification in Sleep Medicine. Two participants who completed the initial assessment and exercise program declined to complete the final PSG assessment because of COVID-19 concerns. Their final sleep EEGs were recorded by Dreem 2, which is comparable to PSG to acquire physiological sleep signals [38].