Last year, we presented concrete use cases in diabetes and cancer, developed within DataSpace4Health, a Gaia-X flagship initiative, that brings together hospitals, research organisations, public institutions and industry partners In Luxembourg. These use cases demonstrated how trusted and federated data environments can enable cross-organisational and cross-border use of health data, and established a practical foundation for trustworthy data sharing in healthcare.

This session builds on that experience and looks ahead to AID-CARE, a newly starting international initiative led from Luxembourg, involving healthcare providers, academic centres and industry partners across Europe and Japan. Core contributors include the Luxembourg Institute of Health, Hôpitaux Robert Schuman, and Japanese NTT Data, working together to address dementia and neurodegenerative disease through data and AI. This domain relies heavily on unstructured clinical information, longitudinal patient trajectories, and multidisciplinary judgement, making it a demanding test case for turning shared data into clinically trustworthy insights and AI-supported decision-making.

As the project is at an early stage, the talk does not present results. Instead, it focuses on the problem space health systems now face: transforming unstructured records into usable data, aligning interoperability models such as OMOP and HL7 FHIR with real clinical workflows, involving clinicians in validation, and embedding governance so that trust is built into data and AI processes from the start. The session is aimed at healthcare leaders and data programme owners who need to understand what “trustworthy data for trustworthy AI” means in practice, beyond pilots and policy statements.