|
Activity Number:
|
220
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
| Abstract - #305124 |
|
Title:
|
Regularized Estimation in AFT Models with High-Dimensional Covariates
|
|
Author(s):
|
Liping Huang*+ and Mai Zhou and Arne C. Bathke
|
|
Companies:
|
University of Kentucky and University of Kentucky and University of Kentucky
|
|
Address:
|
Department of Statistics, Patterson Office Tower, Lexington, KY, 40506-0027,
|
|
Keywords:
|
AFT ; Inverse Probability Weighing ; Glm Net
|
|
Abstract:
|
Several recent researches have focused on the use of AFT models to predict survival times of future cancer patients by investigating their gene expression profiles based on microarray analysis. We investigate use of the elastic net regularization approach to estimation and variable selection in the accelerated failure time model with high-dimensional covariates based on the so called inverse probability of censor weighting method. Huang, Ma and Xie (2006) studied a similar setting using LASSO. However, after ordering the survival times, if the last patient is censored, the current weighting method for that patient is problematic especially so for data with high censoring. We propose and investigate some modified weighting methods for the last patient in this study and show that some modification has improved the prediction performance.
|
