Xianoyan (Iris) Lin

Department of Statistics

University of South Carolina


A Semiparametric Probit Model for Interval-Censored Failure Time Data

The analysis of time-to-event data, generally called survival analysis, arises in many fields of study, including medicine, biology, engineering, public health, epidemiology, and economics. One important feature of survival data is that the failure time of interest is incomplete due to some censoring mechanism. Interval-censored data is a special type of survival data containing left, interval, and right-censored observations. Due to its complex data structure and censoring mechanism, many conventional methods including counting process and martingale techniques, which are successfully used for right-censored data, are not applicable. In this talk, I will present a semi-parametric probit model for analyzing general interval-censored data. We propose to model the unknown non-decreasing function with monotone splines, and develop both Bayesian and likelihood methods under the model. The proposed Bayesian method relies on a novel data augmentation based on the interval-censored data structure and is very easy to implement. Both Bayesian and likelihood methods work very well in the simulation studies and data applications.

The ideas used in the proposed approaches can be generalized to other semi-parametric models for analyzing interval-censored data. I will briefly present some ongoing projects on such extensions.


Back to Colloquium Series