Scottish Regional Safe Havens receive funding to support real-world sensitive data research

Funded as part of DARE UK Phase 2, the projects will run from April 2026 to March 2027, applying new Trusted Research Environment (TRE) capabilities assembled through TREvolution and tested by early adopter projects to answer complex research questions.

In total, eight projects have been awarded funding, with four of these including involvement from Scotland’s Regional Safe Havens which – alongside Research Data Scotland (RDS) and eDRIS at Public Health Scotland – form the Scottish Safe Haven Network. This new funding continues the Network’s success and impact with DARE UK, having already led Sprint and Driver projects in DARE UK Phase 1, and leading the TREvolution transformational programme.

Dr Christian Cole, Academic Co-Director of the Health Informatics Centre (HIC), said: "We welcome this DARE UK funding to support real-world data research. We’re particularly pleased to start operationalising the improvements to TRE infrastructure that HIC is helping to develop as part of the TREvolution programme.

"Alongside our colleagues across the Scottish Safe Haven Network, we’re proud to be involved in four of the eight funded projects, demonstrating Scotland’s role as a leader in UK data infrastructure – a role built on a depth of skills, expertise and experience across Scotland. Ultimately benefiting patients and the public through better secure use of sensitive data in research at scale."

Stuart Dunbar, Engagement Manager at DataLoch, said: "We are delighted to be awarded funding by DARE UK for HEAL-Scot and TransPECT, which will involve ever closer collaboration with partners across the Scottish Safe Haven Network and beyond.

"The unique scope and objectives of these projects underscore the diverse requirements of the TRE ecosystem, and highlight the need for new developments to strengthen the foundations for innovative research that will drive meaningful societal impact in trustworthy ways."

Jen Muir, Scottish Safe Haven Network Manager, said: "RDS is proud to lead the Scottish Safe Haven Network and to see the impact of Scottish expertise on the wider UK data system. Being awarded four out of eight projects really demonstrates the innovation and leadership present in the Scottish research data ecosystem, and the drive to develop and improve our services for public benefit."

Funded projects

The four projects with Scottish Regional Safe Haven involvement are:

HEAL-Scot: Housing, Environment and Location Data Linkage in Federated TREs to Map Geospatial Inequalities

Led by DataLoch in Edinburgh and with involvement from the Health Informatics Centre (HIC) in Dundee, HEAL-Scot will demonstrate how health data can be securely linked with housing, environmental and geographic datasets across multiple Scottish Safe Havens. The project will build federated pipelines connecting DataLoch and HIC, enabling research into how living conditions, environment and location influence health outcomes and inequalities. Find out more

MELODY: Federated Machine Learning for Dermatology

HIC will lead this research project, which will develop and test federated machine learning approaches for dermatology research using clinical images from NHS Tayside and Oxford University Hospitals, with additional support from the Thames Valley and Surrey Secure Data Environment. By training AI models across multiple TREs without centralising data, the project aims to support the development of more inclusive and representative dermatology AI systems. Find out more

SAFEVID: Spasms Analysis using Federated Learning from Videos Across Multiple TREs

Led by the University of Glasgow, with project partners HIC and the West of Scotland Innovation Hub/West of Scotland Safe Haven, SAFEVID will explore how federated machine learning can support early detection of neurological conditions by analysing infant movement videos. By enabling models to learn across multiple TRE datasets without sharing raw video data, the project aims to improve the prediction of conditions such as infantile epileptic spasms. Find out more

TransPECT – Securing AI NLP-Transformer Models for Safe Release in TREs

The TransPECT project is led by DataLoch, with involvement from HIC, Public Health Scotland and King’s College London. The project aims to address the challenge of safely exporting trained AI models from TREs. Focusing on transformer-based language models trained on sensitive free-text data, the project will develop methods and tools to evaluate disclosure risks and ensure that models can be safely released to enable further development and validation work. Find out more

Four further projects have also received funding:

• CONNECT-AF
• FRAME
• BRAID
• AIRR-BRIDGE

Find out more about all the supported projects