Presenter:
Dr. Haocheng Li
Title:
The Statistical Applications of Functional Data Analysis in Physical Activity Studies
Abstract:
Research about the links between physical activity patterns and health outcomes has traditionally been based on fairly crude self-report instruments such as questionnaires. This field is being revolutionized by the availability of relatively inexpensive wearable accelerometer devices. These devices produce over 80,000 data points, second by second, per person per day, and depending on the device, can measure whether the person is asleep, lying down, sitting, standing, and moving, in addition to producing data about the amount of energy expended by physical activity. Statistically, the problem can be cast as densely sampled high dimensional functional data with binary and continuous outcomes. We describe hierarchical multi-dimensional methods that exploit such functional data to show how lifestyle intervention can affect sedentary time, interruptions of sedentary behavior and expenditure of energy from moderate to vigorous physical activity.