What we learnt from the DARE UK Phase 1 Sprint Exemplar Projects
In January 2022, after an open call for applications, we awarded nine research teams funding from UK Research and Innovation (UKRI), totalling more than £2 million, to complete a series of Sprint Exemplar Projects from January to August 2022. These projects aimed to identify and evaluate early thinking for the creation of a coordinated and trustworthy national data research infrastructure to support UK-wide, cross-domain sensitive data research for public benefit.
The DARE UK Phase 1 Sprint Exemplar Projects assessed best practices for the governance, ethical, and public involvement and engagement components involved in establishing a more coordinated data research infrastructure while exploring use cases showcasing the technology. The nine projects were implemented for a period of eight months from January to August 2022, each spanning multiple study fields and UKRI research council mandates.
By the end of the delivery cycle, the projects had successfully demonstrated the benefits of a cloud-based federation approach, which simplifies secure connectivity across a network of trusted research environments (TREs) and equalises access to computational resources while enhancing audit and management reporting. However, the successes of the projects were not without challenges and valuable lessons to take forward.
To get a clear understanding of these challenges and learnings, we embarked on an evaluation exercise with support from management consulting firm Carnall Farrar. Additionally, we conducted a parallel assessment focusing specifically on the Public Involvement and Engagement (PIE) components of the projects. This effort was led by Dr Ester Bellavia, a Public Involvement and Engagement Officer at Health Data Research UK (HDR UK), with support from the former DARE UK Senior Communications and Engagement Manager, Elizabeth Waind, and the then DARE UK Intern, Zahra Atta.
Here’s a summary of our key learnings and reflections from the nine Sprint Exemplar Projects:
- The Sprint Exemplar use cases are a compelling testament to the immense scientific potential that federation can unlock. By uniting research environments, enabling seamless connectivity, and promoting collaboration, federation empowers researchers to push the boundaries of knowledge and achieve breakthroughs that can positively impact society.
- Different use cases require varied solutions for federated analysis. Common Data Models can support federated analytics, but challenges remain, especially for cross-domain data analysis. A federated ecosystem of TREs should enable both federated analytics (where data stays put and queries/results move) and data pooling (bringing data together temporarily if needed).
- Using an ‘Infrastructure as code’ model will help establish an open-source framework for connecting TREs and make it easier for new TREs to join a federated network. By developing and adopting ready-made technologies, we can enhance compatibility and cooperation between different TRE systems, which is crucial for successful TRE federation.
- There is a growing need for community consensus on the minimum standards or reference architecture for TREs – a recommendation being explored by the SATRE project, funded as part of the DARE UK Phase 1 Driver Projects.
- Cloud technology offers equal access to scalable computational resources and simplifies auditing in a federated network. Public cloud platforms enable flexible deployment of resources across the network, while detailed metadata logging enhances audit and management reporting.
- A standardised risk assessment framework enables consistent governance decisions on risk management. It provides explicit criteria for assessing risk and increasing transparency and public trust. Using a standardised approach accelerates decision-making for data owners, facilitating timely project approvals for research.
- Progress is being made in assessing privacy risks from novel research outputs like Machine Learning (ML) models. While traditional statistical analyses on sensitive data are well understood, the potential for attackers to identify individuals through trained ML models remains largely unknown. The DARE UK-funded SACRO Driver Project is looking to address this by developing tools for semi-automated output checking.
We also identified the following key lessons from a PIE perspective:
- PIE provides accessible information, better understanding, and opportunities to respond to public enquiries.
- Insufficient PIE experience and guidance and limited resources negatively impact PIE activities.
- Sharing of lessons learned and the impact of public contributions builds trust and demonstrates commitment.
- Research team buy-in and collaboration with community groups are crucial for successful PIE.
- Ongoing attitude, active listening, and clarity of expectations are important for retaining public members.
- Providing materials in advance and using various formats enhance understanding and responsiveness.
- Online platforms and group chats facilitate remote collaboration and ongoing support.
- Public-led activities or professional facilitation enhance engagement.
- Working with a limited time frame affects meaningful activities, but adequate support and well-defined plans help.
- Diversity, partnering with charities, and consent acquisition are considerations for inclusion.
- Adaptations may be necessary for visual impairments or different levels of knowledge.
- Inclusive PIE practice requires understanding needs and negotiation.
- Resources and costs, including reimbursement, vary and can impact future opportunities.
- Standardised approaches can support consistency and clarity. A PIE playbook has since been developed to guide DARE UK projects going forward.
These reflections and learnings have been vital in informing the direction of the programme, specifically in the delivery of the DARE UK Phase 1 Driver Projects, which have made significant progress in building on the findings from the Sprint Exemplar Projects.
Follow the progress of the DARE UK Phase 1 Driver Projects
Read the full reports:
- DARE UK Phase 1 Sprint Exemplar Projects: Reflections and Lessons Learned
- Public Involvement and Engagement in Sensitive Data Research: Lessons Learned from the DARE UK Sprint Exemplar Projects
