Parametric Changepoint Survival Model with Application to Coronary Artery Bypass Graft Surgery Data
Is there a period of a higher risk of death immediately after a coronary artery bypass? When is the change point? How big is the change in risk of death before and after the change point?
Typical survival analyses treat the time to failure as a response and use parametric models, such as the Weibull or log-normal, or non-parametric methods, such as the Cox proportional analysis, to estimate survivor functions and investigate the effect of covariates. In some circumstances, for example where treatment is harsh, the empirical survivor curve appears segmented with steep initial descent followed by a plateau or less sharp decline. This is the case in the analysis of survival experience after coronary artery bypass surgery, the application which motivated this project. We employ a parametric Weibull change point model for the analysis of such data, and bootstrap procedures for estimation of standard errors. In addition, we consider the effect on the analyses of rounding of the data, with such rounding leading to large numbers of ties.
This type of interdisciplinary work is a hallmark of our program in Applied Statistics at Simon Fraser University. For more information, please contact Suman Jiwani (email@example.com) or her supervisor Charmaine Dean (firstname.lastname@example.org), Department of Statistics and Actuarial Science.