In the future, metadata standards and the explainability of results obtained with artificial intelligence technologies will both play a decisive role to bridge the gap between the data silos of different research disciplines. Yet, data security and data protection run counter to a complete, open exchange of data. This is especially true for the field of medicine. CASUS focuses on technologies and solutions for secure, intelligent, and sustainable data management in digital health. The first application scenario addressed by CASUS is to help medical professionals in making the optimal decisions in cancer treatment.

Patient data analysis using artificial intelligence requires new techniques. For example, metadata management and explainability of the artificial intelligence used are required to preserve the information that is important for complete data provenance management in interdisciplinary research. When it comes to implementation, CASUS is pursuing the idea of making machine learning in digital health possible via a federated and scalable approach. In this way, analysis, model building, and knowledge extraction can be realized close to the patient data.

While storing data centrally results in a huge dataset, the federated approach leaves most of the data where they are. Anonymization or at least pseudonymization of the data is not required, in contrast to central data storage. By bringing the analytics algorithms to the data (and not the other way around), data privacy is assured simply by virtue of the design of the process. CASUS has taken the lead in developing an open-source solution for federated machine learning and artificial intelligence in digital health. This approach will allow artificial intelligence models to be trained on patient data from different sites and the results to be consolidated into a single model.