Bayesian Cox model with categorical predictors for time to event breast cancer data.

Leo Alexander, T.; Pari Dayal, L.; Ponnuraja, C.; Venkatesan, P.

Indian Journal of Applied Research; 2014; 4; 497-501.

Abstract: Survival analysis has become a standard tool for modeling cancer trial data when the event of interest is “time to event” . Cox regression, which implements the proportional hazards model, is designed for analysis of time until an event or time between events, introduced by Cox (1972) in order to estimate the effects of different covariates influencing the time-to-event data. This model has been used extensively in time to event of cancer trial data for given categorical predictor variables. The Bayesian analysis has advantageous in dealing with small sample of censored data more than a frequentist method. The main objective of this article is to apply a Bayesian Cox model and is being compared with a frequentist method. Gibbs sampling technique is used to assess the posterior quantities of interest and to avoid the complexity in calculations. The posterior is arrived using SAS package.


Keywords: Survival data, Cox PH model, Bayesian approach, Gibbs sampler

Back to List of publications / Home