The SeRP team recently attended the International Population Data Linkage Network (IPDLN) Conference in Chicago, USA. Founded in 2007, IPDLN’s mission is to facilitate the exchange of ideas and methods about data linkage and has a global membership of nearly 2,000 researchers and practitioners.
Together with colleagues and partners from Population Data Science at Swansea University, Digital Health Cooperative Research Centre (Australia), and ICES (Canada), SeRP co-led the workshop ‘An international workshop for innovation in synthetic data’. The session explored the relationship between synthetic data and common data models (CDMs), such as the Observational Medical Outcomes Partnership (OMOP) CDM. Workshop participants explored what the relationship between synthetic data and CDMs could offer in terms of enhancing data interoperability, scalability, and privacy preservation within healthcare research, education and innovation (REI) systems.
The workshop aimed to provide a comprehensive understanding of the influential factors affecting the adoption, implementation, and use of synthetic data, CDMs and federation globally within healthcare REI contexts. The workshop also promoted collaboration among relevant stakeholders to explore governance structures, facilitate knowledge-sharing, and support capacity-building initiatives.
Setting the scene.
The workshop began by setting the scene to discuss and identify the key influencers, challenges, enablers and some examples of best practices that are shaping the adoption of synthetic data, CDMs and federation globally. This introduction included brief updates focussed on current initiatives and success stories in advancing this field.
Stakeholder mapping.
A stakeholder mapping exercise aimed to bring Trusted Research Environment (TRE) providers, researchers, educators, and policymakers together to identify common goals, address differences in perspectives, and foster collaboration and alignment. This session also explored stakeholder perspectives to understand needs, expectations, concerns, and motivations.
Establishing Use Cases.
The workshop then turned towards establishing use cases for synthetic data, CDMs and federated data analysis by addressing the following questions:
- What, where when should we use synthetic, CDMs and federated data analysis?
- When is the cost of creating and maintaining CDMs and infrastructure for federated analysis justified within the context of REI?
- Is it possible to develop governance and standards to enable federated data sharing using synthetic data and/or common data models?
Next steps.
Finally, workshop participants looked for opportunities to identify cross-country proof of concept projects that develop operational models, gather feedback from stakeholders and refine the existing approaches. Designed to be an iterative process, this session identified best practices and corresponding challenges or limitations encountered during implementation.
Workshop Outputs.
Throughout the workshop discourse, participant feedback and outputs were captured and shared to support international advancements following the workshop. The needs, challenges and benefits of synthetic data, CDMs and federated analysis were systematically examined from the perspective of:
- Data Custodians/Owners.
- Policymakers.
- Consumers and projects/users requiring or currently accessing data.
- Researchers and those generating intelligence from research.
- TRE providers.
Common cross-cutting themes were extracted to summarise the workshop outputs and were applied to a sharpened focus on the future direction of these innovations, technology requirements for education and research, and potential impacts across stakeholder groups. These valuable perspectives demonstrated how synthetic data can be effectively leveraged to drive innovation and address critical challenges in the healthcare sector.
Reflecting on the workshop, Co-director of SeRP and SAIL Databank, Professor Simon Thompson, said,
“To have so many international contributors, who sit at the forefront of these innovations, all come together at this workshop, was really exciting. It’s vital that we learn and work together to identify knowledge gaps and shared goals to accelerate progress. This will help drive innovations in the use of data and digital technologies like artificial intelligence and large language models to maximise data utility. We’ve seen this emerging in some industrial sectors. Healthcare data is particular problematic, given its necessarily strict privacy controls, but through a shared, collective effort I believe we can achieve it whilst still adhering to the highest privacy and governance standards.”
Professor of Population Data Science Research at Swansea University, Professor Ashley Akbari added,
“It was fantastic to hear from attendees who joined us at the workshop from around the world, who contributed their experiences across synthetic data, common data models and federation. There were great discussions around some of the challenges being experienced across a range of stakeholders and backgrounds, with a range of opportunities for further innovation and collaboration.”
Professor James Boyd, Chair of Digital Health and Innovation at La Trobe University’s Department of Public Health, said,
“Our recent workshop at the IPDLN conference highlighted the synergy between synthetic data and common data models. It was clear from the workshop that by combining these powerful tools, we can accelerate healthcare research, enhance data privacy, and foster international collaboration.”
CEO of ICES Canada, Dr Michael Schull, commented,
“The workshop revealed that our centers are facing many common challenges in implementing common data models and federated analysis, but also that we can advance further and faster, and fulfill the potential of these tools and techniques, by working together on solutions.”