A random horizon model for sequential clinical trials

Abstract

In response to a recent critique of a sequential clinical trials model, a version of the model is described for which the optimal stopping rule is explicitly soluble. The paper seeks to further the dialogue between practitioners and theoretician by describing, with the use of concrete formulas and graphics, a wide variety of the model's operating characteristics. Some should be of particular interest to practitioners and others should be of interest to theoreticians.

Keywords:

• Bayesian
• clinical trials
• frequentist
• Markov chain
• martingale