Bayesian methods for the design and analysis of complex follow-up studies

My doctoral research focused on two topics: i) models for the analysis of multi-state time-to-event data; and ii) decision-theoretic approaches for the design of clinical trials with a survival endpoint. For the first, I developed stochastic processes useful for the Bayesian non-parametric analysis of follow-up studies where patients may experience multiple events relevant to their prognosis. For the second, I developed an approach that uses data from early clinical trials to specify the statistical test used in a confirmatory survival study, accounting for the possible failure of standard assumptions. In this thesis, I describe 3 research papers that report my contributions. Part of my work has been conducted while a visiting researcher at the Dana-Farber Cancer Institute, Boston, Massachusetts (United States of America).