Advances in technology have resulted in the use of sensors in a variety of applications ranging from weather forecasting, GPS tracking to physical activity measurement. Novel analytic techniques need to be developed to model these densely sampled high-dimensional data. Our project focused on approaches to model accelerometry data. Accelerometers measure minute-level human movement, hence provide a rich framework for assessing physical activity patterns of an individual. Using accelerometer data collected in research studies in the School of Medicine at UC San Diego, we applied functional data methods to analyze these Big Data (1440 min * 7-14 days * 550+ individuals). We developed multilevel functional principal component models to ascertain physical activity patterns incorporating temporal and subject-to-subject variation. We then tested if these patterns were associated with health outcomes such as obesity, cancer status, biomarkers and quality of life using regression models. We anticipate that our mathematical framework will provide new statistical and computational tools to study accelerometry and inform societal guidelines on leading a healthy lifestyle.