The aim of this thesis, which was motivated by methodological problems that arise in the context of experimental trials, is to propose a hybrid Bayesian-frequentist approach using sequential Bayesian prediction of the index ...