ABSTRACT
As pointed out by Woodcock (2005), Bayesian approaches to clinical trials are of great interest in the medical product development community because they offer a way to gain valid information in a manner that is potentially more parsimonious of time, resources and investigational subjects than our current methods. The need to streamline the product development process without sacrificing important information has become increasingly apparent. Temper (2005) also indicated that FDA’s reviewers have already used some of the thinking processes that involve Bayesian approaches, although the Bayesian approaches are not implemented. In a Bayesian paradigm, initial beliefs concerning a parameter of interest (discrete or continuous) are expressed by a prior distribution. Evidence from further data is then modeled by a likelihood function for the parameter. The normalized product of the prior and the likelihood forms a socalled posterior distribution. Based on the posterior distribution, conclusions regarding the parameter of interest can then be drawn. The possible use of Bayesian methods in clinical trials have been studied extensively in the literature in recent years. See, for example, Brophy and Joseph (1995), Lilford and Braunholtz (1996), Berry and Stangl (1996), Gelman, Carlin and Rubin (2003), Spiegelhalter, Abrams, and Myles (2004), Goodman (1999, 2005), Louis (2005), Berry (2005), and Berry et al. (2011).
