The Data Sciences Group at PRAIRIE is a dedicated team of experts specializing in the field of machine learning and its applications in health sciences and medical imaging. Their primary goal is to advance the theoretical and computational aspects of these domains to tackle complex challenges. The group focuses on developing innovative computational methods that draw upon various fields such as non-convex optimization, optimal transport, Bayesian statistics, and deformable shapes. By integrating these diverse techniques, they aim to create powerful tools that can effectively address the unique demands of health sciences and medical imaging.
On the theoretical front, the group dedicates significant efforts to providing rigorous theoretical guarantees for the convergence of their optimization schemes and the reliability of their sampling strategies. Through an in-depth mathematical modeling and analysis, they derive theoretical bounds and guarantees that provide insights into the behavior and performance of their methods. This ensures that their computational methods are not only effective but also grounded in solid mathematical foundations.
They actively develop efficient computational packages, which are distributed as open-source Python packages. By making their tools openly available, they promote collaboration and encourage the adoption of their approaches across the scientific community. The group’s numerical schemes find extensive application in various areas of medicine and health sciences. For instance, they contribute to statistical estimation problems, including the monitoring of cancer treatments and the analysis of single-cell genomics data. Furthermore, the Data Sciences Group develop state-of-the-art methods for medical imaging problems. One notable area is shape registration for computational anatomy, where they develop advanced algorithms to align and compare anatomical structures accurately.