Department of Biostatistics

People

Prospective Students

Academics

Research

News & Events

Calendar

Consulting

Employment Opportunities

Resource Quicklinks

Computing Environment

Contact

THESIS DEFENSE ABSTRACT

Cross-Sectional and Longitudinal Penalized Functional Regression

A. Jeffrey Goldsmith, PhD Candidate, Johns Hopkins Department of Biostatistics

Methods for the analysis of cross-sectionally and longitudinally observed images and functions are developed. A tube-fitting algorithm estimates the external boundary and internal properties of anatomical and functional structures in image data. Motivated by tract profiles observed in a neuroimaging context, cross-sectional and then longitudinal functional regression models are proposed, and estimation methods built on the mixed-model framework are explored. Variational Bayes methods, a fast alternative to fully Bayesian computations, are implemented in the functional regression setting, allowing fast and accurate joint modeling of functional predictors and outcomes. Finally, full three-dimensional images are used as predictors of scalar outcomes.