112
reads

Chiung-Yu Huang
2010-02-26  12:30 - 14:30
Room 405, Mathematics Research Center Building (ori. New Math. Bldg.)

The gap times between recurrent events are often of primary interest in medical and epidemiology studies. The observed gap times, however, cannot be naively treated as clustered survival data in analysis because of the unique ordering structure of recurrent events. In this talk, I will introduces two important building blocks, the averaged counting process and the averaged at-risk process, that are useful for the development of estimation and model checking methods for the Cox model. We demonstrate that with the use of these two empirical processes, existing risk-set based methods for univariate survival time data can be easily extended to analyze recurrent gap times. Additionally, we propose a modified within-cluster resampling (WCR) method which can be easily implemented in standard software. We show that the modified WCR estimators are asymptotically equivalent to the risk-set based estimators. I will also describe a class of graphical and numerical testing techniques based on averaging the martingale-like residual processes within a subject. This maneuver is very general and suitable for various purposes of model fitting assessment. An analysis of hospitalization data from a psychiatric register will be presented to illustrate the proposed methods. (This is joint work with Xianghua Luo and Dean Follmann.)