Multivariate failure data received much attention by many researchers, to name a few Akaike (1974), Schwarz (1978), Volinsky and Raftery (2000), Fan and Li (2000, 2001, 2002). The basic assumption in Cox’s proportion hazard model is that the survival time of subjects are independent. This assumption may be violated some time and the collected data may exhibit the existence of correlation among the survival times of the chosen subjects. One popular approach to model correlated survival times is to use a frailty model. Unlike the Cox regression model, there are some challenges in parameter estimation in the Cox frailty model even without the task of model selection. When the correlation among the observations is not of interest, the marginal proportional hazard models have received much attention in the recent literature because they are semi-parametric models and retain the virtue of the Cox model. In this paper, the extension of the Cox regression model to the analysis of multivariate survival time data include Frailty and Marginal hazard models are discussed. Detailed illustrations are also provided.
Impact Factor
2017: 6.614
This work is licensed under a