Rhino

Federated computing architectures enable organizations or teams of organizations to analyze large, multi-source data (‘federated analytics’) or to train machine learning models (‘federated learning’) on such data without compromising the privacy of individual datasets therein, or, if needed, of the model or calculation itself. Rhino Federated Computing provides a fully managed federated computing platform, eliminating the need for enterprise or research customers to build this infrastructure from scratch.

Federated computing is an extremely useful strategy when the data in question is medical data, and associated with natural persons whose privacy must be protected even while their information is mined to advance research in the field. The federated architecture implemented for PROBE will allow data to remain on the server and behind the firewall of its origin institution, while still contributing value in the form of abstracted quantities and parameters to computing efforts driven by researchers or data scientists operating on any network node of the consortium. Deploying and operationalizing this infrastructure, augmented by integrated data harmonization functions, forms the core of Work Package 3. We will work together during this work package with leaders at Uni Hamburg to build the data and computing backbone of PROBE. Downstream, in Work Packages 4 and 5, our informatics and data science experts will cooperate with contributing investigators to ensure that the platform is performing as needed to facilitate their research. In work package 6, we will contribute insights from our enterprise-scale experience to shape PROBE into a sustainable, long-term return on investment.