Covariate Adjusted Distributions of Random Curves

• Activity Number: 79
• Type: Contributed
• Date/Time: Sunday, July 29, 2012 : 4:00 PM to 5:50 PM
• Sponsor: ENAR
• Author(s): Meng Li*+ and Ana-Maria Staicu and Howard Bondell
• Companies: North Carolina State University and North Carolina State University and North Carolina State University
• Keywords: Skewed functional data ; DTI study ; Gaussian and t-copulas ; Covariate analysis

Abstract

The objective of this paper is to propose a new modeling framework for the distribution of skewed curves that accommodates covariate information. The approach extends the ideas of Staicu, Crainiceanu, Reich and Ruppert (2010) by allowing the mean functions to depend on the covariate. However our models are general and will be relevant to many functional data sets, where the aim is to model the functions by accounting for covariate information, especially when the data exhibits pointwise skewness after conditioning on the covariates. We present simulation studies to illustrate the numerical performance of our approach in finite settings. In addition, the proposed method will be developed based on a computationally efficient estimation procedure. Our methodology is motivated by an application to the DTI tractography study to model different modalities of DTI subject profiles, while accounting for the subjects disease status (multiple sclerosis or healthy controls) and their age.