HEREDITARY partners met on 12–13 November 2026 at the Department of Information Engineering of the University of Padua (UNIPD), Italy, for the project’s 8th Federated Learning Workshop. Over two days of technical discussions and joint work, project partners focused on one goal: preparing an integrated live demonstration of federated computation across HEREDITARY’s components, to be presented at the project review in March 2027.
Bringing federated analytics and federated learning together
The planned demonstration will follow a clinical question related to Amyotrophic Lateral Sclerosis (ALS), showing how federated analytics and federated learning can work together across the HEREDITARY Data Network (HDN). Using a visual analytics application developed within WP5, clinicians will be able to compare patient groups, explore their clinical characteristics, and visualise predictions from a survival model trained across different data nodes.
During the workshop, partners worked on defining the queries, data structures and technical connections needed for this end-to-end process. The initial implementation uses ALS data from the BRAINTEASER dataset, distributed between the Turin and Padua nodes. The network is expected to expand to include additional datasets, such as Answer ALS, enabling federated analysis and learning across four nodes by February 2027.
Partners also discussed the integration of PRISM as the HDN privacy layer and progress on synthetic data, developed jointly by Aalborg University and UNIPD. The workshop concluded with a shared roadmap of tasks and responsibilities to bring together the different project developments ahead of the March review.
Understanding federated analytics: A shared language for health data that cannot travel
The workshop also connects with a broader topic explored by HEREDITARY coordinator Gianmaria Silvello, Full Professor of Computer Engineering at the University of Padua, in the article “Federated analytics: A shared language for health data that cannot travel”.
The article explains how federated analytics can help researchers work with health data distributed across hospitals and research centres without having to transfer patient records to a central location. It also highlights the importance of shared data semantics and the role of HERO, HEREDITARY’s multimodal ontology, in ensuring that different institutions can interpret and analyse information consistently.
Through practical examples related to ALS research, Silvello explores how these technologies can support activities such as identifying suitable participants for clinical trials and developing predictive models using clinical, genomic and imaging data from multiple centres.
Read the full article: Federated analytics: A shared language for health data that cannot travel.
Next steps
Work towards the March 2027 demonstration will continue over the coming months, with a follow-up technical workshop planned for January 2027, ahead of the next HEREDITARY plenary meeting. Partners will use this opportunity to further test the integrated pipeline and prepare the live demonstration for the project review.



Recent Comments