Variable Selection for Case-Cohort Studies with Failure Time Outcome

• Activity Number: 126
• Type: Contributed
• Date/Time: Monday, August 4, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Andy Ni*+ and Jianwen Cai
• Companies: and University of North Carolina at Chapel Hill
• Keywords: Variable Selection ; Diverging Parameters ; SCAD ; Oracle Property ; Case-Cohort Design ; Survival Analysis

Abstract

Case-cohort design is widely used in large cohort studies to reduce the cost of covariate measurement. Many of those studies have large number of covariates, especially with the increasing availability of massive genetic information. Therefore, an efficient variable selection method is needed for case-cohort design. In this paper, we study the properties of the Smoothly Clipped Absolute Deviation (SCAD) penalty based variable selection procedure in case-cohort design with failure time outcome. We establish the consistency and asymptotic normality of the maximum penalized pseudo-partial likelihood estimator. We also show that the proposed model selection procedure can identify the true model with probability one as sample size goes to infinity, and it is as efficiently as if the true model is known a priori. Extensive simulation studies are conducted to assess and compare the finite sample performance of the proposed variable selection procedure with AIC- and BIC-based tuning parameter selection methods. We make recommendations for practical use of the variable selection procedures in case-cohort studies. The proposed procedure is applied to the Busselton Health Study.