“…Gamma process is one of the most widely used non-parametric process priors to model the cumulative baseline hazard in Bayesian proportional hazards models (Kalbfleisch, 1978;Clayton, 1991;Sinha et al, 1999Sinha et al, , 2003Sinha et al, , 2015. It has been adopted in many applications including multivariate methods (Dunson and Chen, 2004;Sen et al, 2010;Cai, 2010;Sreedevi and Sankaran, 2012;Ouyang et al, 2013) and variable selection methods (Lee et al, 2011;Savitsky et al, 2011;Gu et al, 2013;Zhao et al, 2014;Lee et al, 2015a), and other survival models (Gelfand and Kottas, 2003;Cho et al, 2009;Zhao et al, 2015). In the absence of the prior information, in the survival models where Gamma process prior is used for cumulative baseline hazard functions (Burridge, 1981;Lee et al, 2011;Savitsky et al, 2011;Gu et al, 2013;Zhao et al, 2014;Lee et al, 2015a), the time partition is specified either based on uniquely ordered failure times or equi-length intervals conditional on the prespecified number of interval cutpoints.…”