Functional Data Analysis in Physical Activity Studies
We describe hierarchical multi-dimensional methods that exploit functional data to show how lifestyle intervention can affect physical activity factors such as sedentary time, interruptions of sedentary behavior and energy expenditure. The regression structures are specified as smooth curves measured at various time-points with random effects that have a hierarchical correlation structure. The random curves for each variable are summarized using a few important principal components. The methods are applied to physical activity data measured by wearable accelerometer devices.