A0280
Title: A general joint latent class model of longitudinal and survival data with covariance modelling
Authors: Ruoyu Miao - University of Cambridge (United Kingdom) [presenting]
Christiana Charalambous - University of Manchester (United Kingdom)
Abstract: Based on the proposed time-varying joint latent class model (JLCM), the heterogeneous random covariance matrix can also be considered, which regression submodel for the variance-covariance matrix of the multivariate latent random effects can be added to the joint latent class model. A general JLCM with heterogeneous random-effects modelling is a natural extension of the time-varying JLCM, which consists of the linear and the log link functions to model the covariance matrices as the variance-covariance regression submodel based on the modified Cholesky decomposition, longitudinal submodel, survival submodel as well as the membership probability. The covariance modelling enables us to determine any effects of covariates on the association between the longitudinal and survival processes while also allowing each subject's group classification to change over time. The assumption can also be tested by adding the regression model and the homogeneous random effects. The Bayesian approach will be used to do the estimation. DIC value is the criteria to decide the optimal k value. Our general JLCM is illustrated on a real data set of aids study in which the prospective accuracy of our proposed JLCM is of interest, as did the dynamic predictions for time-to-death in the joint model using the longitudinal CD4 cell count measurements.
