Building a Global Research Ecosystem: DPUK’s Journey into Federated Data Access
By Emma Squires, Head of Programme Delivery, Dementias Platform UK
Dementia is a global health challenge that affects millions of people and their families. The research community has made significant advances over the last decade, but progress is still constrained by one fundamental issue: access to the right data, at the right time and in a form that can be analysed responsibly and effectively. Much of the most valuable data in dementia research is held within secure environments and across different countries and institutions. It reflects the contributions of participants who have donated their time and personal information to support scientific progress. Safeguarding this data is paramount, but this must not mean limiting the potential impact it can have.
At Dementias Platform UK (DPUK), we have been working to create a research ecosystem where data remains protected, yet collaboration is enabled in a meaningful and scalable way. This work builds on our core principle that scientific discovery moves faster when researchers are able to share methods, ask questions across datasets, and combine insights across different research domains.
Federation: Turning an Ambition into a Deliverable System
The concept of federated data access has gained increasing attention in recent years. The idea is straightforward. Instead of sharing or transferring data across borders or organisations, researchers send analytical workflows to where the data is stored. The data does not move. The insights do.
This approach preserves governance requirements, protects participant privacy and supports compliance with national and institutional data controls. Yet, despite its promise, federation has often been more of a theoretical aspiration than a practical reality.
This changed when DPUK implemented TRE-FX federated solution in partnership with the Alzheimer’s Disease Data Initiative (ADDI). TRE-FX connects DPUK’s secure research environment with the AD Workbench, enabling international researchers to discover and analyse DPUK’s cohort data alongside other dementia datasets. It does this without replicating, exporting or centralising the data and without compromising the governance frameworks that are essential to maintaining trust.
This development has shown that federation does not require compromise. It can deliver both protection and access at the same time.
Demonstrating Value Through Real Research Activity
Since TRE-FX went live, we have seen a strong and diverse set of applications from research groups around the world. These projects span a broad range of scientific priorities which reflects the complexity of dementia research. They include:
- The use of artificial intelligence to support early detection and diagnosis
- Applications of deep learning techniques in neuroimaging analysis
- Multi-omics research designed to explore biological mechanisms
- Harmonisation studies aimed at improving reproducibility across datasets
- Longitudinal modelling to understand disease progression over time
Each of these studies benefits from access to deeply characterised cohort data that has been collected with care and rigour. They are also strengthened by being carried out in a federated environment, where collaboration can occur without weakening data protections.
This move toward federated research is already enabling faster hypothesis testing, more transparent workflows and more robust comparisons between datasets. The early outcomes demonstrate that this model can accelerate discovery while preserving the values we hold as a research community.
Expanding Capability: The Next Phase of Work
The deployment of TRE-FX is an important milestone but it is also the beginning of a broader transformation. We are now preparing the next phase which will focus on expanding the types of research that can take place across federated environments and making these methods more accessible to researchers.
Our priorities include:
- Developing synthetic datasets:
These will allow researchers to explore data structures, test code and develop new analytical approaches before they apply them to sensitive real-world data. - Building federated AI analysis pipelines:
This work will support training and inference across distributed datasets, enabling machine learning models to benefit from larger and more varied data without the need to combine or transfer it. - Creating shared governance frameworks:
By harmonising governance approaches across institutions and countries, we can reduce duplication of effort, improve transparency and create clearer pathways for responsible data access. - Supporting multimodal data linkage:
We are integrating clinical, imaging, genetic, cognitive and digital measures so researchers can investigate dementia from multiple angles within federated environments.
Taken together, these developments will help build a research ecosystem that is scalable, internationally interoperable, and prepared for the
Looking Ahead
The work we are undertaking at DPUK is not only about technology. It is also about designing systems that respect the contribution of research participants and uphold the responsibility that comes with working with sensitive data. Federation is not a shortcut or a workaround. It is a thoughtful and principled model that enables collaboration while preserving the rights and expectations of those who provide the data.
We are still in the early stages of building this new ecosystem but the foundations are now firmly in place. The progress made so far demonstrates that responsible global collaboration is achievable, and that it can accelerate scientific discovery in a way that is ethical, secure and sustainable.
The future of dementia research will depend on our ability to work together across disciplines, borders and organisations. At DPUK we are committed to making that future possible.