| Abstract [eng] |
Increasing number of clinical trials contributed to the increased need of analysis of censored survival data. Cox proportional hazards model has become a great success in this area, which was published in 1972 (Cox, 1972). Regression analysis of survival data has become possible. This model was proposed in order to evaluate effects of the different covariates, which influence the time to event (survival) data. It is discussed PHM right-censored data for mortality basing on the Bayesian approach. The main aspect of the Bayes rule is prediction. Improvement of the new numerical algorithms, such as Gibbs sampling allowing obtain a sample from posterior value, motivated to start using the Bayesian methods in the survival analysis. It is selected baseline hazard simulation using the polygonal function for the assessment of survival curves in order to avoid inaccurate and misleading results. This paper is intended to show primarily the importance of including the covariates to the prediction of survival distribution. Also it is presented review of Bayesian MCMC method for Cox model and described the baseline hazard simulation by using the polygonal function. Finally, it is showed how this model can be adapted in using the software WinBUGS. Method is illustrated with examined practical tasks. |
