A Class of Semiparametric Transformation Models for Doubly Censored Failure Time Data

• Published in: Scandinavian journal of statistics Vol. 45; no. 3; pp. 682 - 698
• Main Authors: LI, SHUWEI, HU, TAO, WANG, PEIJIE, SUN, JIANGUO
• Format: Journal Article
• Language: English
• Published: Oxford Wiley Publishing 01.09.2018; Blackwell Publishing Ltd
• Subjects: Asymptotic properties; Computer simulation; doubly censored data; Economic models; Epidemiology; Independent variables; Inference; Maximum likelihood estimation; nonparametric maximum likelihood estimation; Nonparametric statistics; Regression analysis; semiparametric efficiency; transformation models; Transformations
• ISSN: 0303-6898; 1467-9469
• DOI: 10.1111/sjos.12319

Doubly censored failure time data occur in many areas including demographical studies, epidemiology studies, medical studies and tumorigenicity experiments, and correspondingly some inference procedures have been developed in the literature (Biometrika, 91, 2004, 277; Comput. Statist. Data Anal., 57, 2013, 41; J. Comput. Graph. Statist., 13, 2004, 123). In this paper, we discuss regression analysis of such data under a class of flexible semiparametric transformation models, which includes some commonly used models for doubly censored data as special cases. For inference, the non-parametric maximum likelihood estimation will be developed and in particular, we will present a novel expectation–maximization algorithm with the use of subject-specific independent Poisson variables. In addition, the asymptotic properties of the proposed estimators are established and an extensive simulation study suggests that the proposed methodology works well for practical situations. The method is applied to an AIDS study.