Dynamic prediction for multiple repeated measures and event time data

Jue Wang, The University of Texas School of Public Health

Abstract

In many clinical trials studying neurodegenerative diseases such as Parkinson's disease (PD) and amyotrophic lateral sclerosis (ALS), multiple longitudinal outcomes are collected to fully explore the multidimensional impairment caused by these diseases. In the first part of this work, we propose a joint model that consists of a semiparametric latent trait linear mixed model (LTLMM) for the multiple longitudinal outcomes, and a survival model for event time. The two submodels are linked together by an underlying latent variable. We develop a Bayesian approach for parameter estimation and a dynamic prediction framework for predicting target patients' future outcome trajectories and risk of a survival event, based on their multivariate longitudinal measurements. Our proposed model is evaluated by simulation studies and is applied to the DATATOP study, a motivating clinical trial assessing the effect of deprenyl among patients with early PD. In the second part of this work, we relax the unidimensional assumption of the LTLMM model and propose a multidimensional latent trait linear mixed model (MLTLMM) that allows multiple latent variables and within-item multidimensionality (one outcome can be a manifestation of more than one latent variable). We conduct extensive simulations studies to assess and compare the performance of unidimensional and multidimensional latent trait models under various scenarios. The simulation studies suggest that the multidimensional model outperforms unidimensional model when the multivariate longitudinal outcomes are related to multiple latent variables. The proposed model is applied to the motivating clinical trial of ceftriaxone in subjects with ALS and the Pooled Resource Open-Access ALS Clinical Trials (PRO-ACT) database. In the third part, we extend the proposed MLTLMM model to the joint modeling framework by introducing a Cox’s proportional hazards model with piecewise constant baseline hazard for event time data. Our proposed model is evaluated by simulation studies and is applied to the Ceftriaxone study.

Subject Area

Biostatistics|Statistics

Recommended Citation

Wang, Jue, "Dynamic prediction for multiple repeated measures and event time data" (2016). Texas Medical Center Dissertations (via ProQuest). AAI10179084.
https://digitalcommons.library.tmc.edu/dissertations/AAI10179084

Share

COinS