Functional Data Analysis in Digital Health

BST 268 Subject & Catalog Number

Description

New technologies are beginning to reshape scientific questions and priorities in public health and medical research. Smartphones and wearables can now provide nearly continuous data for weeks or months or even years, enabling precise measurement of outcomes, covariates, and exposures. Examples of emerging data types include accelerometer data on physical activity, GPS data on mobility, continuous glucose monitoring data, and ecological momentary assessments. Functional data analysis (FDA) refers to a suite of statistical techniques intended for settings where the underlying data generating mechanism is continuous in nature. Rather than thinking of a single observation as consisting of one (or more) numbers as the outcome and one (or more) numbers as the predictor, the outcomes and/or predictors can be thought of as being functions of time or space. The intended audience of this course is master’s and PhD students in biostatistics as well as all students interested in learning about methods for such data.