Research Project: Assessment Methodologies in High Dimensional Problems

This project addresses issues in the validation of the efficacy of devices that reduce high dimensional data, usually for diagnostic purposes. The primary goals of this project are to assess which study designs and statistics are appropriate and most informative in evaluating efficacy, and, more importantly, methods to estimate bias and uncertainty in those studies and statistics, even in cases where optimal designs are not available. Often these statistics are nonparametric rank-sum statistics used in diagnostic settings.