The project is aimed at analyzing nonrandom dropout in longitudinal brain imaging studies, performed at Umeå center for Functional Brain Imaging (UFBI). The images are large (about 100000 image elements, voxels, or more per subject). The analysis accounting for nonrandom dropout involves Bayesian methods which in turn require Monte Carlo methods for their implementation. Usually, 1000-10000 Monte Carlo iterations are required for convergence, and in each iteration draws from high-dimensional multivariate distributions are performed. This is very computationally demanding.
Short background about the brain imaging applications of interest for the project: Measuring the brain through different imaging techniques provide important insights on the various aspects of brain structure and function and how they are related to the cognitive aging process. Life expectancy of the world population is increasing, and as a consequence the number of individuals with cognitive impairments related to aging is expected to double within a 50-year period. In order to really follow the process of aging, on an individual basis, we must follow people over time and measure them repeatedly, i.e. perform a longitudinal study. Unfortunately, not all individuals that enter our studies see them through to the end, we have dropout. It is fair to assume that the dropout is related to e.g. the brain imaging outcome, in which case the dropout is called non-ignorable and require special statistical considerations when analyzing the collected data.