|
Activity Number:
|
104
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #309573 |
|
Title:
|
Joint Analysis of Multiple Longitudinal Outcomes and a Time-to-Event Using a Nonlinear Latent Class Approach
|
|
Author(s):
|
Cecile Proust-Lima*+ and Helene Jacqmin-Gadda and Jeremy Taylor
|
|
Companies:
|
University of Michigan and INSERM, U875 and University of Michigan
|
|
Address:
|
Department of Biostatistics SPH, Ann Arbor, MI, 48109-2029,
|
|
Keywords:
|
Mixed model ; conditional independence assumption ; joint analysis ; maximum likelihood ; latent class ; longitudinal data
|
|
Abstract:
|
We propose a joint model for exploring the association between correlated longitudinal non-Gaussian markers and a time-to-event. A longitudinal latent class model describes the different latent classes of evolution of the latent process underlying the markers, and a proportional hazard model describes the risk of event according to the latent classes. The latent process is linked to the quantitative non-Gaussian markers by using nonlinear transformations which include parameters to be estimated. Depending on whether the function of risk is a parametric or semi-parametric function, a maximum likelihood estimation or a penalized likelihood approach is used. We propose a posterior residual analysis to evaluate the conditional independence hypothesis given the latent classes and apply the methodology in the context of cognitive aging.
|
