CFE 2019: Start Registration
View Submission - CMStatistics
Title: Joint modeling and estimation for multivariate longitudinal data and binary outcome Authors:  Toshihiro Misumi - Yokohama City University (Japan) [presenting]
Hidetoshi Matsui - Shiga University (Japan)
Abstract: In medical research areas, a joint modeling that simultaneously incorporates multivariate longitudinal biomarker processes and a binary outcome process has attracted considerable attention. The joint model consists of a multivariate linear mixed effects model and a logistic regression model with shared random effects. Numerous unknown parameters included in the two submodels are simultaneously estimated by joint maximum likelihood method. We discuss the effective estimation procedure based on a pseudo-likelihood and a h-likelihood. The estimated joint model provides a powerful tool to know how closely the multivariate longitudinal trajectories of biomarkers are associated with a clinical outcome. We also discuss the relationship between the joint modeling and functional data modeling. Some numerical studies are conducted to investigate the effectiveness of our proposed modeling strategy.