%20(12).png)
.png)
The TRE-FX project aimed to address the intricate challenges of executing data analysis across multiple Trusted Research Environments (TREs) with diverse geographical and governance boundaries. Bitfount platform acted as the link within the TRE ecosystem, to enable analyses involving the training and deployment of AI models. This enables researchers to perform analysis on secure data within the TREs in a way that is constrained, auditable, and safe.
The Challenge of Multiple TREs
Trusted Research Environments (TREs) are secure locations in which data is placed for researchers to analyse. However, it is hard for a researcher to perform analysis across multiple TREs. The TRE-FX project was designed to tackle the complexities researchers face when working across multiple TREs, particularly those with different regional and governance structures. The project's primary objective was to streamline this process, making it easier, safer, and more efficient for researchers to access and analyse data.
Technical Innovations
As part of the TRE-FX project, Bitfount helped develop and implement key technologies:
Five Safes RO-Crate
This new Digital Object format was created to standardise the flow of requests, results, and metadata between federated analysis platforms and TREs for workflow-based computational analysis. A compliant crate encapsulates data workflows and provenance in a machine and human-readable package, ensuring the safety and appropriateness of data access and analysis.
Component-Driven Architecture
The project established a component-driven architecture using approved workflows for query execution within TREs. This open-source architecture facilitated the execution of queries and enhanced data security and governance compliance. The architecture is modular and contains 4 layers:
Submission Layer: This layer allows researchers to securely submit their task execution (TES) requests linked to the Five Safes RO-Crate. It acts as the initial interface between users and the TREs, handling requests and updates via a secure API.
TRE Controller Layer: Serving as an intermediary, this layer manages the transition of requests from the external world to the internal TRE processing environment. It ensures policy compliance and prepares requests for processing by unpacking and storing the RO-Crate.
Workflow Executor Layer: Positioned within the TRE, this layer’s primary function is to process the RO-Crate, execute the requested analyses, and manage the results. It plays a critical role in ensuring data security and integrity during analysis.
Transparency Layer: An innovative addition, this layer enhances research oversight and auditability. It publishes a simplified version of the Five Safes RO-Crate and the associated workflows, providing an external view of the analytical tasks and metadata for public access and review
Workflow Execution Integration
The project utilised the ELIXIR Workflow Execution System (WfExS) and integrated it to operate in closed environments. This integration aligned TRE-FX with various European programmes and enhanced the project's interoperability and flexibility.
Enhanced Provenance Model
The Five Safes RO-Crate profile includes a detailed provenance model, crucial for tracking and validating TRE operations, ensuring compliance with data security policies, and providing an audit trail.
Bitfount’s Adaptation and Integration
Bitfount adapted its federated data science platform to implement the Five Safes RO-Crate and integrated its technology with the TRE-FX architecture. This adaptation included modifying the submission layer to dispatch Five Safes RO-Crates and utilizing the Bitfount Pod as the TRE-Controller. This implementation demonstrated Bitfount's ability to maintain interoperability within the TRE-FX framework and its commitment to flexible, secure federated analytics.
Impact and Future Directions
This project demonstrated the seamless integration capability of data federation across TREs with third-party entities, demonstrating how researchers can safely combine data from many sources, and how data providers from any sector can safely implement this using technology and standards we already have today.
The TRE-FX project's success extends beyond technical achievements. It has significantly influenced the design of the next phase of the DARE UK programme and contributed to the creation of a streamlined, collaborative, and secure data analytic environment, applicable across various domains. The project's outcomes are also being incorporated into major European initiatives, demonstrating the scalability and adaptability of the technologies and frameworks developed.
Bitfount is committed to fostering an open, secure, and collaborative research environment while unlocking the benefits of sensitive data. To learn more about Bitfount's role in the TRE-FX project and how our solutions can support your federated data science and AI initiatives, please contact us.
Read the TRE-FX team's final project report here: https://zenodo.org/records/10084398
TRE-FX was funded by UK Research and Innovation as part of Phase 1 of the DARE UK (Data and Analytics Research Environments UK) programme, delivered in partnership with Health Data Research UK (HDR UK) and ADR UK (Administrative Data Research UK). In collaboration with ELIXIR-UK, ELIXIR, The University of Manchester, University of Nottingham, University of Dundee, Swansea University, University of Liverpool, University of Birmingham and University Hospitals Birmingham NHS Foundation Trust.
.png)
.png)
