Bayesian cure rate mixture model for time to event breast cancer patients data

Abstract

This study utilizes Bayesian cure rate mixture models to analyze breast cancer patient data through several survival distributions, including Weibull, Lognormal, Exponential, and Gompertz with the Gamma distribution serving as a prior. We computed a variety of postulated models across multiple iterations to derive posterior statistics for each distribution. Model convergence was assessed using the Deviance Information Criterion (DIC), allowing us to determine the most suitable distribution for our data. Our findings indicate that the lognormal distribution emerged as the best-fitting model for the breast cancer patient data, highlighting its effectiveness for this application.