Title: Bayesian semiparametric inference and selection for dynamic treatment regimes
Authors: David Stephens - McGill University (Canada) [presenting]
Abstract: Computational strategies are developed that allow fully Bayesian inference to be carried out for possibly misspecified models. Behaviour under, and robustness to, misspecification is widely studied in the frequentist world but is not prominent amongst Bayesians. We will demonstrate how the computational approaches provide exact Bayesian inference and how this can be deployed in the setting of dynamic treatment regimes.