A1289
Title: Reproducible learning for accelerated failure time models via deep knockoffs
Authors: Daoji Li - California State University Fullerton (United States) [presenting]
Abstract: Selecting truly relevant variables contributing to the response is a fundamental problem in many scientific fields. One of the major challenges in variable selection is effectively controlling the false discovery rate (FDR). Most existing variable selection procedures in survival analysis neglect the FDR control. Such a gap is filled, and a new and flexible variable selection method with guaranteed FDR control is proposed for accelerated failure time models. The proposed method combines the strengths of deep knockoffs and the weighted M-estimation procedure and enjoys the FDR control for arbitrarily high dimensions with finite samples. More importantly, the proposed method does not require prior knowledge about the joint distribution of covariates. Extensive simulation studies confirm the proposed method's generality, effectiveness, and power. Finally, the proposed method is used to analyze primary biliary cirrhosis data to demonstrate its practical utility.
