Speaker: Erjia Cui, PhD Candidate in Department of Biostatistics, Johns Hopkins University
Title: Wearable computing in biobank-scale research: a functional data perspective
Abstract: Wearable devices have been increasingly deployed in large epidemiological and clinical studies to provide objective measures of human activity in the free-living environment. The vast amount of objectively measured physical activity data collected by these wearable devices is gradually reshaping physical activity research. However, analyzing wearable data in biobank-scale studies is challenging because of its size, dimensionality, and complexity. I will introduce some recent advances in functional data analysis (FDA) methods for wearable computing, including: (1) survival analysis of functional predictors with nonlinear structure; and (2) fast, scalable approaches for regression and variability decomposition of complex-structured functional data. Methods will be illustrated using the accelerometry data from the National Health and Nutrition Examination Survey (NHANES). The results of these analyses provide novel findings on the association between objectively measured physical activity and health outcomes and characteristics of study participants. I will also briefly discuss two ongoing projects based on UK Biobank data studying the association between physical activity and two different health outcomes: multiple sclerosis and breast cancer.
Note: Seminars are free and open to the public. Reception to follow.
