A Test of Fit for The Generalized Koziol-Green Model with Recurrent Event Data

摘要

The generalized Koziol-Green model is used for analyzing recurrent event data where the right censoring random variable is informative of the inter-event time distribution. We propose a test of fit for this model based on a weighted difference of two estimators of the cumulative hazard rate function. Asymptotic properties of this difference are given. Test statistics of the chi-squared, Kolmogorov-Smirnov and Cramer-von Mises form are constructed based on these results. A real data set example is used to illustrate these tests.