Spring Seminar Series - March 2, 2006
University of Minnesota
School of Statistics
College of Liberal Arts

Joint Modelling of Survival and Longitudinal Data

Jane-Ling Wang
Department of Statistics
University of California at Davis

Thursday, March 2, 2006
3:30 PM, 115 Ford Hall
Minneapolis, East Bank Campus
Social at 3:00 PM, 300 Ford Hall

Abstract

In many longitudinal studies the event-times of interest (termed failure times) are collected along with baseline and longitudinal covariates. Relationship between a failure time process and some longitudinal covariates is of key interest, so is the understanding of the patterns of the longitudinal covariate processes. Several difficulties arise when covariates are measured intermittently, possibly with error, and no measurements are available after the event-time, which triggers nonignorable dropout. Semiparametric joint likelihood approaches have emerged as effective ways to model both processes. However, there are challenges both on the computational front and for statistical inference. In this talk, we address both issues, which are intertwined, and discuss several practical solutions.

Typically, a parametric longitudinal model is assumed to facilitate the likelihood approach. However, the choice of a proper parametric model turns out more illusive than standard longitudinal studies where no survival end-point occurs. Furthermore, the computational burden due to both Monte Carlo numerical integration and EM (Expected Maximum) algorithm is an important concern in the joint modelling setting. To deal with those challenges, in the first part of the talk, we will explore several nonparametric longitudinal models in the joint modelling setting. In the second part of the talk, we will introduce the method of sieves for joint modelling to illustrate the high dimensionality problem currently encountered in the joint modelling literature. The asymptotic properties of the proposed sieve estimator will be discussed.