On Bayesian estimation of the multiple decrement function in the competing risks problem, II: The discrete case

Abstract

In an earlier paper, the authors studied the behavior of Bayes estimators of the multiple decrement function in the competing risks problem using a Dirichlet process prior with a continuous prior measure α. Certain applications, including the analysis of life tables based on grouped data, require a complementary analysis with a discrete α. Such an analysis is developed here, and applied to provide Bayes estimators of the multiple decrement function from discrete data on the survival times of multiple myeloma patients.