Properties of estimators of baseline hazard functions in a semiparametric cure model |
| |
Authors: | Tomoyuki Sugimoto Toshimitsu Hamasaki |
| |
Institution: | (1) Department of Biomedical Statistics, Graduate School of Medicine, Osaka University, 2-2 Yamada-oka, Suita, Osaka 565-0871, Japan |
| |
Abstract: | We consider a semiparametric cure model combining the Cox model with the logistic model. There are the two distinct methods
for estimating the nonparametric baseline hazard function of the model; one is based on a pseudo partial likelihood and the
other is to use an EM algorithm. In this paper, we discuss the consistency and the asymptotic normality of the estimators
from the two methods. Then, we show that the estimator from the pseudo partial likelihood can be characterized by the (forward)
Volterra integral equation, and the estimator from the EM algorithm by the Fredholm integral equation. These characterizations
reveal differences in the properties between the estimators from the two methods. In addition, a simulation study is performed
to numerically confirm the results in several finite samples. |
| |
Keywords: | Cox model Logistic model Nonparametric baseline hazard Volterra integral equation Fredholm integral equation Asymptotic property |
本文献已被 SpringerLink 等数据库收录! |
|