|
Activity Number:
|
400
|
|
Type:
|
Invited
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #307779 |
|
Title:
|
Semiparametric Approaches for the Analysis of Multilevel Failure Time Data
|
|
Author(s):
|
Joanna H. Shih*+ and Shou-En Lu
|
|
Companies:
|
National Cancer Institute and University of Medicine and Dentistry of New Jersey
|
|
Address:
|
6130 Executive Blvd, EPN room 8132, Bethesda, MD, 20892-7434,
|
|
Keywords:
|
multi-level clustering ; nested frailties ; Monte-Carlo EM ; Within-cluster resampling
|
|
Abstract:
|
Multilevel clustered failure time data arise when clustering of data occurs at more than one level. Often it is of interest to make inference on the association of failure times at each level of clustering. Two modeling approaches (i.e., marginal and conditional) are considered for analyzing this type of data. For the marginal approach, we consider a class of multivariate survival models parameterized by marginal distributions and accounting for hierarchical structure of clustering through copula functions. This class of models emphasizes on a population-averaged interpretation for the covariate effects on the marginal hazard. For the conditional approach, we reparameterize the above class of models such that they are represented as nested random effects proportional hazard models.
|
