|
Activity Number:
|
196
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, July 30, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #309268 |
|
Title:
|
Inverse Regression Estimation for Censored Data
|
|
Author(s):
|
Nivedita Nadkarni*+ and Michael Kosorok
|
|
Companies:
|
University of Wisconsin-Madison and The University of North Carolina at Chapel Hill
|
|
Address:
|
1521 E Franklin Street, Chapel Hill, NC, 27514,
|
|
Keywords:
|
right censored data ; sufficient dimension reduction ; inverse probability of censored weighting ; inverse regression
|
|
Abstract:
|
An objective of analyzing survival data via regression is to develop a predictive model given covariates. An important step in formulating the model involves variable selection. Selection of the influential predictors is critical and becomes complicated if the data has high dimensional covariates, as is often the case in many clinical trials and more recently microarray studies. In addition to selection, assessment of predictor performance is also crucial. In this paper, we develop sufficient dimension reduction methodology via inverse regression for censored data. An inverse probability of censoring weighted approach is implemented to facilitate variable selection. A simulation study and several data analyses demonstrate the effectiveness of the proposed procedure.
|
