Semi-parametric regression analysis of interval-censored failure time data||Semi-parametric regression analysis of interval-censored failure time data

This dissertation discusses the regression analysis of interval-censored data which includes the current status data as a special case. In the first part of this dissertation, we adopt the linear transformation models for regression analysis of interval-censored data and propose an empirical likelihood-based procedure to address the problem of underestimating variance of estimated parameters. In the second part of this dissertation, we will focus on the situation when the covariates of some subjects could be missing or cannot be measured exactly but there exist some auxiliary covariates. We propose an estimated partial likelihood approach for estimation of regression parameters that make use of the available auxiliary information. In the third part and fourth part of this dissertation, we will discuss the regression analysis of current status data and interval-censored data, respectively, when the censoring mechanism could be related to the failure time of interest. The copula model and monotone I-splines are used in these two parts and asymptotic properties of the resulting estimates are established in both cases. Also simulation studies are conducted in each of the four parts to evaluate the finite sample properties of the estimators. Illustrative examples are also provided.