DataTools4Heart_Milestone MS14_Secure federated network for cardiology health data set up in 7 European sites built using N > 1,000,000 EHRs and compliant with GDPR, and EHDS roadmap

This document reports on the achievement of Milestone MS14 of the DataTools4Heart (DT4H) project, which concerns the establishment of a secure federated network for cardiology health data across European clinical sites, compliant with GDPR and aligned with the European Health Data Space (EHDS) roadmap. As of March 2026, the DT4H platform has been deployed in its V3.0 release, representing the first production-ready version of the federated learning infrastructure. The platform is organised into three functional layers (Resources and Community, Data Management, and Federated Processing), providing consortium partners with tools for dataset discovery, metadata exploration, AI-assisted data querying, and the configuration and execution of federated learning experiments across sites. A detailed description of the platform modules and their current operational status is provided in Section 1. Section 2 presents the deployment roadmap for the remainder of the project. Following the V3.0 release, the platform will enter a beta testing phase (April–July 2026) with real clinical data, leading to a feature-complete V4.0 deployment by the end of August 2026, and culminating in the mature platform final release at the end of September 2026. Section 3 addresses the data justification for the milestone. All eight clinical partners have carried out heart failure (HF) cohort extractions, contributing a combined total of approximately 150,000 patients from centres across the United Kingdom, Spain, Sweden, Romania, Czech Republic, Italy, and The Netherlands. The section discusses why reaching the originally envisioned scale of one million electronic health records (EHRs) presents inherent constraints, including finite patient populations per centre, varying levels of hospital digitisation, and the rigorous application of data minimisation principles under GDPR, while contextualising the achieved dataset size against comparable large-scale HF registries.