ABSTRACT
The methods commonly known as survival analysis address data in the form of time to end-point. This chapter examines a large dataset of prostate cancer from the SEER registry for Louisiana. The Kaplan -Meier (KM) estimator is available in the survival package in R. The KM plot suggests that there is a racial effect in survival with blacks apparently having poorer long term survival. For a Bayesian geospatial analysis people need to consider a likelihood and consider how to incorporate spatial effects within the formulation. A different approach is sometimes taken to modeling survival data. The Weibull distribution is a proportional hazards model and also there are some associated interpretational issues with such a model, especially when mediator effects are to be considered. Instead it is possible to consider a model that directly relates the log of time to explanatory variables or effects. The accelerated failure time model makes some assumptions.
