On the estimators and tests for the semiparametric hazards regression model

详细信息    查看全文

• 作者：Seung-Hwan Lee
• 关键词：Accelerated failure time model ; Accelerated hazards model ; Censored data ; Log ; rank estimating function ; Proportional hazards model
• 刊名：Lifetime Data Analysis
• 出版年：2016
• 出版时间：October 2016
• 年：2016
• 卷：22
• 期：4
• 页码：531-546
• 全文大小：506 KB
• 刊物类别：Mathematics and Statistics
• 刊物主题：Statistics
  Statistics
  Statistics for Life Sciences, Medicine and Health Sciences
  Quality Control, Reliability, Safety and Risk
  Statistics for Business, Economics, Mathematical Finance and Insurance
  Operation Research and Decision Theory
  
• 出版者：Springer Netherlands
• ISSN：1572-9249
• 卷排序：22

文摘

In the accelerated hazards regression model with censored data, estimation of the covariance matrices of the regression parameters is difficult, since it involves the unknown baseline hazard function and its derivative. This paper provides simple but reliable procedures that yield asymptotically normal estimators whose covariance matrices can be easily estimated. A class of weight functions are introduced to result in the estimators whose asymptotic covariance matrices do not involve the derivative of the unknown hazard function. Based on the estimators obtained from different weight functions, some goodness-of-fit tests are constructed to check the adequacy of the accelerated hazards regression model. Numerical simulations show that the estimators and tests perform well. The procedures are illustrated in the real world example of leukemia cancer. For the leukemia cancer data, the issue of interest is a comparison of two groups of patients that had two different kinds of bone marrow transplants. It is found that the difference of the two groups are well described by a time-scale change in hazard functions, i.e., the accelerated hazards model.