Abstract
Sleep and activity patterns have been linked to physical performance in older adults. Traditional parametric models of 24-hour activity rhythms fail to adequately capture specific diurnal sleep and wake patterns; functional principal components analysis (fPCA) is a non-parametric approach that addresses this limitation. Using fPCA, we modeled accelerometry data from n = 2,960 participants in the Osteoporotic Fractures in Men (MrOS) ancillary sleep study (mean age 77y) and examined cross-sectional associations with gait speed and grip strength measurements. Lower daytime activity (expected difference = -0.049 [-0.072, -0.028] m/s), increased sleep duration and a reduced midday dip in activity (-0.015 [-0.035, 0.006] m/s) were modestly associated with worsening gait speed. A modest association between both later sleep and wake times and increased sleep duration with worsening grip strength outcomes was observed (-1.11 [-1.90, -0.32] kg). Specific daily activity patterns may serve as predictive biomarkers for changing physical function in aging populations.