Robust Bayesian Inference for Censored Survival Models

This paper proposes a robust Bayesian accelerated failure time model for censored survival data. We develop a new family of life-time distributions using a scale mixture of the generalized gamma distributions, where we propose a novel super heavy-tailed distribution as a mixing density. We theoretically show that, under some conditions, the proposed method satisfies the full posterior robustness, which guarantees robustness of point estimation as well as uncertainty quantification. For posterior computation, we employ an integral expression of the proposed heavy-tailed distribution to develop an efficient posterior computation algorithm based on the Markov chain Monte Carlo. The performance of the proposed method is illustrated through numerical experiments and real data example.