CMStatistics 2018: Start Registration
View Submission - CMStatistics
Title: Joint correlation rank screening for semi-competing risks data with ultrahigh-dimensional gene features Authors:  Liming Xiang - Nanyang Technological University (Singapore) [presenting]
Mengjiao Peng - Nanyang Technological University (Singapore)
Abstract: Ultrahigh dimensional gene features are often collected in modern cancer studies, where the number of gene features is extremely larger than the sample size. We propose a joint screening procedure for survival data subject to semicompeting risks with ultrahigh dimensional covariates. The method employs the ranking of the correlation between covariates and the joint survival distribution of both nonterminal and terminal event times. It is model-free and easy to implement as it only requires Kaplan-Meier estimators for the joint survival function of both event times. Theoretical properties of the proposed method are established. Simulation results show that the proposed screening procedure works very well for different types of marginal models. The practical utility is illustrated through the analysis of data from a breast cancer study.