scientific article; zbMATH DE number 1462111

From MaRDI portal
Publication:4486028





The basic model, which is the main object of this dissertation, is based on the three following random variables: survival time, censoring time and a vector of covariables. The survival time and the censoring time are often replaced by the observable time and the censoring indicator. The aim of this work is the study of the asymptotic properties of the sieve-maximum-likelihood estimators of the conditional log-hazard function. It is supposed, that this function is the sum of an unspecified function of \(t\) (time) and a linear function of the covariables. Some mathematical instruments of this work correspond to the theory of empirical processes. Two general sieve types are studied. First, the linear models or the linear sieves are considered. Second, the functions which can be represented as finite linear combinations of an infinite set of basis functions are considered. The term ``network is used with regard to medical applications. A sufficient criterion for the \(L^1\)-consistency of estimators for both sieve types is formulated. The consistency for general censored data is the object of a special chapter. The speed of convergence is also studied (the maximal inequality for the log-likelihood functional).











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4486028)