A1052
Title: Robust rank canonical correlation analysis for multivariate survival data
Authors: Di He - Nanjing University (China) [presenting]
Yong Zhou - Chinese Academy of Sciences (China)
Hui Zou - University of Minnesota (United States)
Abstract: Canonical correlation analysis (CCA) is widely applied in statistical analysis of multivariate data to find associations between two sets of multidimensional variables. However, CCA often cannot be used directly for survival data or their monotone transformations, owing to right-censoring in the data. A new robust rank CCA (RRCCA) method is proposed based on Kendall's $\tau$ correlation and adjusts it to deal with multivariate survival data without requiring any model assumptions. Owing to the nature of rank correlation, the RRCCA is invariant against monotone transformations of the data. The estimation consistency of the RRCCA approach is established under weak conditions. Simulation studies demonstrate the superior performance of the RRCCA in terms of estimation accuracy and empirical power. Lastly, the proposed method is demonstrated by applying it to Stanford heart transplant data.
