Classification of Multivariate Longitudinal Data Through a Semiparametric Mixed-Effects Model with Nonstandard Assumptions

• Activity Number: 27
• Type: Contributed
• Date/Time: Sunday, July 31, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Rolando De la Cruz and Cristian Meza*+
• Companies: Pontificia Universidad Católica de Chile and Universidad de Valparaíso
• Keywords: Classification ; Mixed-effects ; Multiple responses ; Splines ; Skew-normal distribution ; EM algorithm

Abstract

Multivariate longitudinal data arise when a set of different responses on the same subject are measured repeatedly over time. This talk discusses model-based statistical methods for the classification of subjects into one of two or more groups based on multivariate longitudinal data. We relate the observed responses using a semiparametric multivariate linear mixed-effects models (SMLMM) to describe evolutions in different groups. The SMLMM is formulated with fixed-effects, random effects and with a smooth function to model the time effect. For each of the response variables, the smooth function is estimated via low-rank cubic splines using radial basis functions. In the SMLMM the error term and the random effects are assumed to follow a multivariate skew-normal distribution. Parameter estimation is done via the expectation-maximization algorithm. Also, we consider the case of missing responses. We present an example using data from a study in 161 pregnant women in Santiago, Chile, where the main interest is to predict normal versus abnormal pregnancy outcomes.