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

Multilevel Joint Analysis of Longitudinal and Binary Outcomes

Hong, Seo Yeon (2013) Multilevel Joint Analysis of Longitudinal and Binary Outcomes. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

This is the latest version of this item.

[img]
Preview
PDF
Primary Text

Download (555kB) | Preview

Abstract

Joint modeling has become a topic of great interest in recent years. The models are simultaneously analyzed using a shared random effect that is common across the two components. While these methods are useful when time-to-event data are available, there are many cases where the outcome of interest is binary and a logistic regression model is used. We propose the use of a joint model with a logistic regression model being used for the binary outcome and a hierarchical mixed effects model being used for the longitudinal outcome. We link the two sub-models using both subject and cluster level random effects and compare it with models using only one level of random effects. We use the Gaussian quadrature technique implemented in the software package aML (Multiprocess Multilevel Modeling software). Simulation studies are presented to illustrate the properties of the proposed model. We also applied our model to the repeated measures of mid-arm muscle circumference (MAMC) and mortality rate for patients within 75 units from 15 centers from a randomized study of hemodialysis (HEMO) and found that the model performs well. We further extend this work by developing methods that can be used to calculate individualized predictions based on our proposed joint model. We use the Bayesian approach to obtain these predictions and implement the method in the software package WinBUGS. The proposed method provides a mechanism for understanding the relationship between a longitudinal measure and a given binary outcome. Thus, it can be used to address several types of public health problems. First, it can be used to understand how changes in a biomarker or other longitudinal measure are related to changes in status of a subject. Second, it can be used to predict the outcome of a subject based on the trajectory of the longitudinal outcome providing information that can be used in a personalized medicine setting. This allows researchers to identify potentially harmful patterns and intervene at an earlier stage.


Share

Citation/Export:
Social Networking:
Share |

Details

Item Type: University of Pittsburgh ETD
Status: Unpublished
Creators/Authors:
CreatorsEmailPitt UsernameORCID
Hong, Seo Yeonseh72@pitt.eduSEH72
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairWeissfeld, Lisalweis@pitt.eduLWEIS
Committee MemberJeong, Jong-HyeonJeong@nsabp.pitt.eduJJEONG
Committee MemberChang, Chung Chouchangj@pitt.eduCHANGJ
Committee MemberRosengart, Matthew Rrosengartmr@upmc.eduMRR18
Date: 30 January 2013
Date Type: Completion
Defense Date: 19 November 2012
Approval Date: 30 January 2013
Submission Date: 30 October 2012
Access Restriction: 2 year -- Restrict access to University of Pittsburgh for a period of 2 years.
Number of Pages: 52
Institution: University of Pittsburgh
Schools and Programs: Graduate School of Public Health > Biostatistics
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: joint model, shared random effect, logistic regression, longitudinal, hierarchical, mixed effects, individual prediction
Date Deposited: 30 Jan 2013 15:09
Last Modified: 19 Dec 2016 14:40
URI: http://d-scholarship.pitt.edu/id/eprint/16754

Available Versions of this Item

  • Multilevel Joint Analysis of Longitudinal and Binary Outcomes. (deposited 30 Jan 2013 15:09) [Currently Displayed]

Metrics

Monthly Views for the past 3 years

Plum Analytics


Actions (login required)

View Item View Item