|
Activity Number:
|
74
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 6, 2006 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #305824 |
|
Title:
|
Semiparametric Analysis of Longitudinal Data with Potential Right Censoring
|
|
Author(s):
|
Mengling Liu*+ and Zhiliang Ying
|
|
Companies:
|
New York University School of Medicine and Columbia University
|
|
Address:
|
650 First Ave., New York, NY, 10016,
|
|
Keywords:
|
counting process ; latent variable ; least square estimation ; marginal model ; normal transformation
|
|
Abstract:
|
A package of semiparametric models is proposed for longitudinal data under possibly irregular observation occasions and potentially informative censorship. The models are motivated by the idea of shared random effects in joint modeling of longitudinal responses and event times, and are valid under a variety of assumptions of censoring mechanism. Specifically, we assume a semiparametric normal transformation model for the informative censoring time and a semiparametric regression model for the longitudinal response variable conditional on the observed censoring information. Asymptotically unbiased estimating equations are constructed and yield least square type estimators for the finite dimensional regression parameters of the semiparametric regression model. The estimators are consistent and asymptotically normal with the variance matrix ready to be estimated by the plug-in rule.
|
