Hazard Change Point Estimation

Abstract

In survival analysis or reliability and maintenance studies, hazard rate function plays an important role, as it describes the instantaneous failure probability given that the individual survived or the item has not failed up to that instance. Sudden changes in the hazard function may occur due to overhauls, treatment effects, or maintenance activities. A problem of statistical interest is the estimation of the times of the changes, as well as their sizes. It is also of interest to study the impact of covariates on the hazard rate function. In the article, we review the basic estimation methods for hazard change point models for complete and censored data. We also discuss extensions that allow for multiple change points and provide a brief exposure to Bayesian approach.

Performance of methods to conduct mediation analysis with time‐to‐event outcomes

“…Further research is needed to fully explain these results. It is, however, important to note that in practice it is uncommon to observe a constant hazard, as there are several factors that might affect the hazard rate during the study (Gürler, ), for example, the length of follow‐up, intrinsic characteristics of a specific health condition, characteristics of treatment, and even selection bias (Hernan, ).…”