Imputation-Based Efficient Estimation Procedures for Length-Biased Data Under Semiparametric Regression Models

• Activity Number: 286
• Type: Topic Contributed
• Date/Time: Tuesday, August 3, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Hao Liu*+ and Jing Qin and Yu Shen
• Companies: Baylor College of Medicine and National Institute of Allergy and Infectious Diseases and MD Anderson Cancer Center
• Keywords: right-censored length-biased data ; Estimating equations ; Efficiency ; semiparametric regression model

Abstract

We consider imputation-based estimation procedures for the regression analysis of right-censored length-biased data under semiparametric transformation models. The procedures jointly estimate the regression coefficients and the baseline hazard function, which can facilitate the prediction of covariate-specific survival probabilities for future patients. New computation algorithms are proposed, and sophistic techniques using the empirical processes method are developed to establish the large-sample properties. The efficiency of the estimation procedures are compared with those of existing methods. Extensive simulation study under small to moderate sample sizes shows that the proposed procedures are more efficient than the existing estimation methods. We demonstrate the estimation procedure by analyzing a real data.