|
Activity Number:
|
433
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 5, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #304463 |
|
Title:
|
Random Effects Semiparametric Regression Model for Clustered Interval-censored Data
|
|
Author(s):
|
K. F. Lam and Yongxian Long*+
|
|
Companies:
|
The University of Hong Kong and The University of Hong Kong
|
|
Address:
|
Rm 518, Meng Wah Complex, Hong Kong, International, , China
|
|
Keywords:
|
Semiparametric frailty model ; Clustered interval-censored data ; Nonlinear-effect covariate ; Unspecified cumulative baseline hazard ; Sieve maximum likelihood estimation ; Piecewise linear
|
|
Abstract:
|
This paper considers a random effects (frailty) semiparametric regression model for the analysis of clustered interval-censored data. Two nonparametric components are involved in the model with the first one being the unspecified cumulative baseline hazard function; while the second one is the nonparametric function of a certain covariate, say age, which has a nonlinear effect on the log-hazard, conditioned on the clustered effect. The distribution of the frailty term can be very flexible as long as the Laplace transform has a close form. Two parametric piecewise linear functions are used to approximate the cumulative baseline hazard function and the nonlinear function. The sieve MLE is proposed and its asymptotic properties will be discussed. Simulation studies suggest that the proposed method works well. The proposed methodology is illustrated by using the data from an influenza study.
|
