Seminars

Marginal regression approach in cure models with clustered discrete survival data

78
reads

Yun-Chan Chi

2010-11-12
12:45:00 - 14:45:00

405 , Mathematics Research Center Building (ori. New Math. Bldg.)

Clustered survival data with a cure fraction arise naturally from biomedicine, econometrics and sociology studies. The mixture cure rate models have been well developed for univariate and multivariate (or clustered) continuous right-censored data. When the correlation structure within clusters is not of interest, Yu and Peng (2008) used a marginal regression approach to construct estimating equations for estimating the parameters in mixture cure rate models. Recently, Zhao and Zhou (2008) proposed two mixture cure rate models for univariate grouped or discrete-time survival data. However, their methodologies can not be directly applied to clustered discrete-time survival data. Therefore, the marginal regression approach is proposed to construct estimating equations based on cure rate models. The accuracy of the estimators is examined by simulation. In addition, the implementation of the marginal approach to a dental implant study is presented.
Keywords : clustered survival data, discrete-time survival data, marginal regression approach, mixture cure rate model