|
Activity Number:
|
433
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
ENAR
|
| Abstract - #303627 |
|
Title:
|
Kronecker Product Linear Exponent AR(1) Correlation Structures for Multivariate Repeated Measures
|
|
Author(s):
|
Sean L. Simpson*+ and Lloyd J. Edwards and Keith E. Muller
|
|
Companies:
|
Wake Forest University and The University of North Carolina at Chapel Hill and University of Florida
|
|
Address:
|
Department of Biostatistical Sciences, , ,
|
|
Keywords:
|
Multivariate repeated measures ; Kronecker product ; Generalized autoregressive model ; Doubly multivariate data ; Spatio-temporal data ; Separable covariance
|
|
Abstract:
|
When analyzing multivariate repeated measures data, it is often advantageous to model the correlation separately for each repeated factor. This method of modeling the correlation utilizes the Kronecker product to combine the factor specific correlation structures into an overall correlation model. We propose a Kronecker product linear exponent autoregressive (LEAR) correlation structure for multivariate repeated measures data in which the correlation between measurements for a given subject is induced by two factors. The model allows for an imbalance in both dimensions across subjects. This four-parameter structure is especially attractive for the High Dimension, Low Sample Size cases so common in medical imaging and various kinds of "-omics" data. We employ the model in the analysis of a longitudinal imaging data example concerning schizophrenia and caudate morphology.
|
