top of page

Orchestration for FAIR-enabled data stations

Updated: May 17

3 DEC 2024


MEETING SUMMARY

FAIR Data Train: The FAIR Data Train introduces a domain-agnostic system that enhances automated interoperability through privacy-focused "data visiting", robust metadata, and scalable, user-centric solutions for FAIR data management.


FAIR Data Train - Specification: A flexible and iterative methodology ensures the FAIR Data Train architecture integrates with existing systems, supporting scalable, practical solutions for diverse stakeholder needs.


HINQ: HINQ repurposes primary healthcare data for both primary and secondary uses, aligning with European frameworks and demonstrating FAIR principles through impactful projects like maternal health risk analysis.


Visma Connect: Visma Connect emphasizes secure, federated data exchange, fostering trust and cross-domain collaboration with scalable, FAIR-compliant solutions for initiatives like LIFES.


DZNE - Swarm Learning: Swarm Learning enables decentralized and secure AI model training, exemplified by projects like the Finnish leukemia study, advancing collaborative AI aligned with FAIR principles.


TNO/CoE: TNO consolidates and scales earlier project outcomes through the Heracles initiative and the Centre of Excellence for Data Sharing and Cloud, funded by the Dutch Ministry of Economic Affairs.


Discussion on Specifications: The discussion focused on enabling interoperability through FAIR metadata, agile development, and leveraging existing systems while ensuring machine-actionable metadata and software components.


General Discussion: Key topics included addressing collaboration challenges, aligning with European frameworks, forming dedicated teams, and prioritizing practical tools and resources for scalable FAIR compliance.



MEETING NOTES




FAIR Data Train specifications


FAIR Data Train - The FAIR Data Train approach was introduced as a domain-agnostic system designed to enhance automated interoperability while adhering to FAIR principles. Key aspects include privacy-focused "data visiting", foundational agreements for metadata handling, and core components like data stations and gateways. The environment promotes user-centric control, robust metadata, and simplified system integration, providing a scalable framework for FAIR data management. The separation of distributed, data holding stations and applications making (station controlled) use of the data is a key concept in the FAIR data train ecosystem.

FAIR Data Train - Specification - A flexible methodology for designing the FAIR Data Train architecture was presented, emphasizing interoperability without replacing existing systems. The development process includes iterative agile cycles, rich documentation, and scalable designs. This approach ensures practical, evolving solutions aligned with the diverse needs of stakeholders and supports seamless updates across domains.




HINQ - Repurposing primary healthcare data for primary use

Hans Niendieker hans.niendieker@hinq.nl & Niels Maijers niels.maijers@hinq.nl

HINQ - HINQ focuses on repurposing primary healthcare data for primary use, but would also aim to make subsets of data available for secondary uses like research and training of machine learning models. Highlighted projects include resolving medication discrepancies and identifying risks in maternal health. The initiative aligns with national and European frameworks and leverages existing infrastructures to demonstrate FAIR principles, setting an example for future projects.




SureSync by Visma - Scalable data exchange and federative data spaces

SureSync by Visma - SureSync shared their expertise in secure, scalable data exchange and federative data spaces, emphasizing system-to-system communication and minimal duplication. Their solutions facilitate FAIR-compliant data sharing across domains and support cross-domain collaboration, trust-building, and user-centric workflows for LIFES initiatives.





DZNE - Swarm learning

Joachim Schultze joachimschultze@s-khb.de


DZNE - Swarm Learning - Swarm Learning technology was showcased as a decentralized method for secure AI model training without transferring data. Real-world applications, like the Finnish leukemia project, demonstrated its scalability and effectiveness. The approach aligns with FAIR principles, enabling collaborative AI and advancing global healthcare data sharing.






Key discussion points

Discussion on Specifications - The main theme of the discussion was a commitment to collaborate with the aim to show that with FAIR metadata and proper (explicitly published) machine actionable metadata and software components, all these existing systems can interoperate without replacement of legacy components or choices, unless these are not compatible with FAIR principles. The discussion focused on defining metadata templates, ensuring interoperability, and testing specifications iteratively. Agile development methods were recommended, emphasizing shared protocols, practical scalability, and leveraging existing systems to achieve FAIR compliance efficiently.

General Discussion - Key topics included collaboration challenges, scaling FAIR principles, and prioritizing practical use cases. Participants agreed on forming dedicated teams, aligning with European frameworks, and leveraging strategic partnerships. Upcoming funding and meetings aim to refine priorities and advance implementation. Practical tools and resources were emphasized to simplify FAIR compliance and data management for organizations and researchers.








Comments

bottom of page