First Hitting Time Regression Models baixar o livro de graça

A useful class of models in the analysis of lifetime (or survival or reliability) data is based on representing the study event as the outcome of an underlying stochastic process which represents, for example, the growth of a flaw in a material or the state of health of a patient. The observed event occurs when the stochastic process reaches a boundary or threshold for the first time. Lifetimes thus correspond to first hitting times (FHT) and allowing them to depend on measured covariates leads to FHT regression or threshold regression models. Models of this type are increasingly finding applications in recent years in recognition of their greater flexibility in comparison to routinely-applied techniques such as proportional hazards models. Methodological developments include the use of a greater variety of underlying stochastic processes (which are most commonly Wiener, Ornstein-Uhlenbeck or Gamma processes), the introduction of additional heterogeneity through individual random effects, models for recurrent events, and the study of the relationship between FHT regression models and the more widely known proportional hazards and accelerated life regression models. The aim of this work is to provide a comprehensive, integrated review of FHT modelling of times-to-event and its place within the general context of the analysis of lifetime data.