Challenges and Possible Solutions for Survival Analysis with High-Dimensional Covariates

• Activity Number: 51
• Type: Invited
• Date/Time: Sunday, August 1, 2010 : 4:00 PM to 5:50 PM
• Sponsor: IMS
• Author(s): Sihai Dave Zhao*+ and Yi Li
• Companies: Harvard University/Dana-Farber Cancer Institute
• Keywords: Variable selection ; high dimensional data analysis ; Cox model

Abstract

Current variable selectors are unable to handle situations where the number of covariates under consideration is ultra-high. Consider a motivating clinical study of multiple myeloma, where progression-free survival and expression levels of more than 25000 genes were measured for each of 170 patients. This dataset defies analysis even with regularized regression. Some remedies have been proposed for the linear model and for generalized linear models, but there are few solutions in the survival setting and, to our knowledge, no theoretical justifications. In this paper we propose a method for handling ultra-high-dimensional covariates in the analysis of censored data and give theoretical support. Simulation studies show that our method performs well even under model misspecification. We apply the proposed procedure to analyze the aforementioned myeloma study.