During the past a few years, part of my statistical methodology research was focused on a very difficult fundamental problem for bivariate survival data analysis: the nonparametric bivariate survival function estimation. The lack of computationally convenient and efficient nonparametric estimator of bivariate survivor function deters the progress in a range of some very important multivariate survival data problems. From this research, my colleagues, Ross Prentice, Zoe Moodie and I have developed a proper, efficient, and computationally convenient nonparametric bivariate survivor function estimator, which provides a basic tool for the display and comparison of censored bivariate survival data, just like Kaplan-Meier estimate for the univariate survivor function estimation. The major research results were summarized in the two papers entitled as “Hazard-Based Nonparametric Survivor Function Estimation” (Prentice, Zoe and Wu, 2004), and “Nonparametric Estimation of the Bivariate Survivor Function: Research Synthesis and Proposals for New Estimators”. These research results have paved the way for developing new statistical methodology in the area of multivariate survival data analysis. The methodology results have potential application to medical study and clinical trial. For instants, disease occurrence studies between pairs of family members in genetic epidemiology.
This article appears in the Journal of Royal Statistical Society Series B 2004. Other authors include Ross Prentice (Fred Hutchinson Cancer Research Center) and Zoe Moodie (Fred Hutchinson Cancer Research Center).
Prentice RL, Moodie FZ, Wu J. Hazard-based nonparametric survivor function estimation. J Royal Stat Soc Series B 66(2):305-319, 2004.