136 – Bayesian Modeling in Physical Sciences and Engineering
Analysis of Survival Data With a Cure Fraction Under Generalized Extreme Value Distribution
Fangfang Liu
Iowa State University
This paper introduces generalized extreme value (GEV) distribution to analyze right-censored survival data for populations with a surviving fraction. Our proposed GEV model leads to extremely
flexible hazard functions. We show that our Bayesian model has several nice properties. For example, we prove that even when improper priors are used, the resulting posterior distribution could
still be proper under some weak conditions. We further provide theoretical and numerical results
showing that our GEV models offer a richer class of models than the widely used Weibull models.