Generalized Odds-Rate Hazards Models for Current Status Data Using EM Algorithm

• Activity Number: 80
• Type: Contributed
• Date/Time: Sunday, August 9, 2015 : 4:00 PM to 5:50 PM
• Sponsor: Biometrics Section
• Author(s): Bin Yao* and Lianming Wang
• Companies: University of South Carolina and University of South Carolina
• Keywords: Current status data ; EM algorithm ; GORH model ; Monotone splines ; Regression analysis

Abstract

Generalized odds-rate hazards (GORH) models, also referred to as G(\rho) family, is a general class of semiparametric regression model taking several popular survival models as special cases, such as the proportional hazards model and proportional odds model. In spite of its generality, few research has been done to analyze current status data or interval-censored data using the GORH model. In this project, we investigate the GORH model for current status data and discuss the non-identifiability issue in the GORH model. Treating \rho as an unknown parameter does not yield valid inference. We propose an efficient expectation-maximization (EM) algorithm for the maximum likelihood estimates when \rho is fixed and known. While the true \rho is unknown, we propose to use a working model with \rho is one for the analysis in which the estimates allow one to conduct statistical inference under the true model. The estimates under the working model are comparable as those under the true model as shown in our simulation study. Our methods are illustrated with real current status data from an epidemiological study on uterine fibroids.