The growing use of health and other sensitive data offers major opportunities to improve large-scale research, but accessing information held by different organisations in disparate repositories presents a significant challenge. Researchers need ways to analyse data across multiple locations without compromising privacy, weakening security or asking organisations to surrender control of their data.
A new paper in the Journal of the American Medical Informatics Association (JAMIA) describes an approach designed to address this challenge: the 5 Safes Task Execution Service (TES). This approach provides a practical way to run approved analyses across different Trusted Research Environments (TREs), also known as Secure Data Environments (SDEs). The research brings together the internationally recognised Five Safes framework, the Global Alliance for Genomics and Health (GA4GH) TES standard, to deliver SeRP’s data federation technology at Swansea University and a tool for submitting data analysis requests between TREs developed at University of Nottingham.
5 Safes Task Execution Service (5 Safes TES)
The paper, Five safes task execution service: a standards-based architecture and reference implementation for secure federated analytics across trusted research environments, describes the design, implementation and evaluation of this system for securely sending analytical tasks to multiple TREs.
Many organisations operate TREs; controlled digital environments where approved researchers can access and analyse sensitive information under strict security and governance arrangements. However, different TREs may use different technologies and operating procedures, making it difficult to conduct consistent research across organisations.
The 5 Safes TES addresses this by standardising how analysis requests are packaged, submitted, executed and returned. Researchers can submit a prepared analytical query to participating TREs, where it runs locally against the data held there. Rather than transferring personal-level records to a central location, the system returns approved results for further analysis.
Who was involved in the research?
The international research team brought together expertise from the University of Nottingham, including Health Informatics, Digital Research Service and NIHR Nottingham Biomedical Research Centre, Swansea University’s Secure eResearch Platform (SeRP) team, the University of Manchester’s Department of Computer Science, and research and government organisations in Singapore, including the Government Technology Agency, the Agency for Science, Technology and Research (A*STAR) and the National Research Foundation. The research was funded by Health Data Research UK.
The role of federation
The paper brings the Five Safes principles and data federation together through a working technical service: 5 Safes TES. Its architecture has three main layers.
The first, the Submission and Orchestration Layer, receives researchers’ analysis requests and coordinates their delivery. The second, the Controller Layer, operates within each TRE and checks requests against local rules, including researcher permissions and project approvals. The third, the Analysis Execution Layer, runs the approved workflow against the data within the secure environment. Outputs then undergo disclosure control before they can be returned to researchers.
This structure is important because it connects the technical process directly to the Five Safes. Standardised requests make analyses more consistent, while local controls preserve each organisation’s authority over its data and the release of results.
Importantly, this approach has already been proven within SeRP tenancy, NDP Dementias Hub (formerly the DPUK Data Portal). In addition, the paper also details a second deployment connecting it with two TREs in Singapore. Using a standardised data format called OMOP CDM, which ensures all clinical information is organised and understood in the same way, the researchers can run identical computational tasks across the UK and Singaporean sites. By utilising the Five Safes framework, the project guarantees that all data access and research activities meet strict security and privacy standards before any analysis begins.
While researchers can submit their analysis requests from afar, the actual processing happens locally inside each system. The UK and Singaporean sites maintain their own controls to check that the requests follow local rules and ensures the analysis is safe before it runs. Once the tasks are completed locally, only the final, aggregated results are sent back to the researchers – not the private data itself.
This approach allows organisations in different countries to work together seamlessly without needing to consolidate their data into one central location, thereby protecting data sovereignty while still enabling sophisticated, large-scale, international research.
Read the full paper here:
https://academic.oup.com/jamia/advance-article/doi/10.1093/jamia/ocag160/8885674
