Comparing first hitting time and proportional hazards regression models

作者：Stogiannis D; Caroni C^*; Anagnostopoulos C E; Toumpoulis I K

来源：Journal of Applied Statistics, 2011, 38(7): 1483-1492.

DOI：10.1080/02664763.2010.505954

• survival analysis
• first hitting time
• inverse Gaussian
• proportional hazards
• Cox regression
• cured fraction

Cox's widely used semi-parametric proportional hazards (PH) regression model places restrictions on the possible shapes of the hazard function. Models based on the first hitting time (FHT) of a stochastic process are among the alternatives and have the attractive feature of being based on a model of the underlying process. We review and compare the PH model and an FHT model based on a Wiener process which leads to an inverse Gaussian (IG) regression model. This particular model can also represent a ocured fractiono or long-term survivors. A case study of survival after coronary artery bypass grafting is used to examine the interpretation of the IG model, especially in relation to covariates that affect both of its parameters.

• 出版日期2011