Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Prediction in the Joint Modeling of Mixed Types of Multivariate Longitudinal Outcomes and a Time-to-Event Outcome

Choi, Jiin (2012) Prediction in the Joint Modeling of Mixed Types of Multivariate Longitudinal Outcomes and a Time-to-Event Outcome. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

This is the latest version of this item.

Primary Text

Download (1MB) | Preview


A common goal of longitudinal studies is to relate a set of repeated observations to a time-to-event endpoint. One example of such a design is in the area of a late-life depression research where repeated measurement of cognitive and functional outcomes can contribute to one's ability to predict whether or not an individual will have a major depressive episode over a period of time. This research proposes a novel model for the relationship between multivariate longitudinal measurements and a time-to-event outcome. The goal of this model is to improve prediction for the time-to-event outcome by considering all longitudinal measurements simultaneously.
In this dissertation, we investigate a joint modeling approach for mixed types of multivariate longitudinal outcomes and a time-to-event outcome using a Bayesian paradigm. For the longitudinal model of continuous and binary outcomes, we formulate multivariate generalized linear mixed models with two types of random effects structures: shared random effects and correlated random effects. For the joint model, the longitudinal outcomes and the time-to-event outcome are assumed to be independent conditional on available covariates and the shared parameters, which are associated with the random effects of the longitudinal outcome processes. A Bayesian method using Markov chain Monte Carlo (MCMC) computed in OpenBUGS is implemented for parameter estimation.
We illustrate the prediction of future event probabilities within a fixed time interval for patients based on our joint model, utilizing baseline data, post-baseline longitudinal measurements, and the time-to-event outcome. Prediction of event or mortality probabilities allows one to intervene clinically when appropriate. Hence, such methods provide a useful public health tool at both the individual and the population levels.
The proposed joint model is applied to data sets on the maintenance therapies in a late-life depression study and the mortality in idiopathic pulmonary fibrosis. The performance of the method is also evaluated in extensive simulation studies.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Choi, Jiinjic31@pitt.eduJIC31
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairAnderson, Stewartsja@pitt.eduSJA
Committee MemberJeong, Jong-Hyeonjeong@nsabp.pitt.eduJJEONG
Committee MemberThompson, Wesley
Committee MemberYouk, Ada O.ayouk@pitt.eduAYOUK
Committee MemberReynolds, Charles F.reynoldscf@upmc.eduCHIPR
Date: 24 September 2012
Date Type: Completion
Defense Date: 8 August 2012
Approval Date: 24 September 2012
Submission Date: 20 July 2012
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 88
Institution: University of Pittsburgh
Schools and Programs: School of Public Health > Biostatistics
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: Joint model; Mixed binary and continuous data; Multivariate longitudinal data; Prediction model; Shared parameter model; Survival analysis
Date Deposited: 24 Sep 2012 18:55
Last Modified: 19 Dec 2016 14:38

Available Versions of this Item

  • Prediction in the Joint Modeling of Mixed Types of Multivariate Longitudinal Outcomes and a Time-to-Event Outcome. (deposited 24 Sep 2012 18:55) [Currently Displayed]


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item