Web-based Software for Real-time Simulation-assisted Trial Design in Alzheimer's Disease.

Abstract

Clinical trials for Alzheimer's Disease (AD) are necessarily designed in the presence of substantial quantitative uncertainty. Certain important aspects of this uncertainty can be mitigated by developing longitudinal models for AD progression and by using these models to simulate virtual trials and estimate operating characteristics (such as statistical power, the probability of stopping at an interim analysis, the probability of identifying the correct dose, etc.) as a function of candidate design features, such as inclusion / exclusion criteria. In this brief report we describe the development and deployment of a customized software solution that allows such simulation-based results to be generated "on the fly" in the context of a drug development team meeting. This solution leverages a number of recent practical advances in statistical and scientific computing that could be much more broadly leveraged to assure more quantitatively grounded trial designs in Alzheimer's Disease.