Recent Developments in Multivariate Survival Analysis

• Activity Number: 51
• Type: Invited
• Date/Time: Sunday, August 1, 2010 : 4:00 PM to 5:50 PM
• Sponsor: IMS
• Author(s): Lei Xu*+ and Ross L. Prentice
• Companies: Fred Hutchinson Cancer Research Center and University of Washington
• Keywords: Bivariate ; Censoring ; Nonparametric ; NPMLE ; Survival data

Abstract

Ross L. Prentice and Lei Xu Fred Hutchinson Cancer Research Center, and University of Washington A brief history will be presented of the challenging bivariate survivor function (BVSF)estimation problem. Available nonparametric estimators include 'plug-in' estimators, which have good moderate sample properties, but incorporate negative mass and are not nonparametric efficient. There are also nonparametric maximum likelhihood(NPML) estimators that use some form of data grouping or smoothing to achieve unique estimation, but moderate sample size properties tend to be poor relative to the plug-in estimators. Here we consider a NPML procedure with specified marginals for unique likelihood maximization. We also combine this with marginal survivor function estimators that can improve upon KM estimators by using information on the companion failure time variable for censored observations.