Seminar by Julia Wrobel

Please Note: This seminar will be given online.

Registration for wearable device data with application to circadian rhythm chronotype discovery

Wearable devices provide a compelling framework for understanding circadian rhythms and diurnal patterns of activity. Using methods we developed for separating amplitude and phase variability in exponential family functional data, we uncover the distinct phenotypes, or chronotypes, that give rise to differences in these patterns in physical activity. Our method for aligning, or registering, observed activity data alternates between two steps: the first uses generalized functional principal components analysis (GFPCA) to calculate template functions, and the second estimates smooth warping functions that map observed curves to templates.

Existing approaches to registration have primarily focused on continuous functional observations, and the few approaches for discrete functional data require a pre-smoothing step; these methods are frequently computationally intensive. In contrast, we focus on the likelihood of the observed data and avoid the need for preprocessing, and we implement both steps of our algorithm in a computationally efficient way. The results of both steps of our algorithm provide unique, data-driven insights into the processes behind magnitude and timing of physical activity. Our motivation comes from the Baltimore Longitudinal Study on Aging, in which accelerometer data provides observations of activity and sedentary behavior.