ESTIMATION AND TESTS IN FINITE MIXTURE-MODELS FOR CENSORED SURVIVAL-DATA

This paper presents a survival model where the hazard function is rela ted to a density that is either a single distribution or a mixture of k parametric distributions. The main purpose of the statistical infere nce is to study the ''susceptibility ratio'' parameter. Using martinga le methods, we prove asymptotic properties of the maximum likelihood e stimates of the model parameters and we test the hypotheses of total c ontamination and nonmixture.