The objective of the multidisciplinary project FLUTE (FEDERATED LEARNING AND MULTI-PARTY COMPUTATION TECHNIQUES FOR PROSTATE CANCER) is to advance and expand data-driven healthcare by developing novel methods for cross-border utilization of data centers while ensuring advanced privacy preservation. Advanced research will be carried out to push the performance boundaries of secure multi-party computation in federated learning, including the associated AI models and secure execution environments. The technical innovations will be integrated into a platform that will enhance privacy and provide innovators with a secure and tested environment for the development, testing, and deployment of federated AI solutions in healthcare, including the integration of real-world health data from data hubs and the generation and use of synthetic data.
To maximize impact, adoption, and replicability of results, the project will contribute to the development of the global standard HL7 FHIR and create new guidelines for cross-border federated learning in healthcare that comply with GDPR.
To demonstrate the practical use and impact of the results, the project will integrate the FLUTE platform with healthcare data hubs located in three different countries, use their data to develop a novel federated AI toolkit for clinically significant prostate cancer diagnosis, and carry out a multinational clinical validation of its effectiveness. This will help improve predictions for aggressive prostate cancer, avoid unnecessary biopsies, thus improving patient well-being and significantly reducing associated costs.
Specifically, within the project, GRADIANT will be responsible for:
Leading WP2: Scalable Privacy-Enhanced Federated Learning and AI.
Researching, designing, implementing, and testing a customized set of PET software/hardware methods adapted to FL.
Developing prediction models for the diagnosis of clinically significant prostate cancer (csPCa) based on MRI clinical imaging data.
