Title: Feature selection with survival outcome data
Authors: Hyokyoung Grace Hong - Michigan State University (United States) [presenting]
Abstract: Detecting biomarkers that are relevant to patients' survival outcome is crucial for precision medicine. Dimension reduction is key in the process. Although regularization methods have been used for dimension reduction, they cannot handle a large number of candidate biomarkers generated by modern bio-techniques. Variable screening, which has been widely used for handling exceedingly large numbers of variables, is however much underdeveloped for censored outcome data. A series of new feature screening procedures for survival data with ultrahigh dimensional covariates is introduced. These methods include conditional screening, integrated powered density (IPOD) screening, $L_q$-norm learning, and forward regression with partial likelihood. We will discuss the intuition behind and demonstrate their utilities through real data analyses.