Australian Health Data Evidence Network (AHDEN): Building a National Data Infrastructure for Standardised, Federated Health Data Research (Infrastructures to Support Data-Intensive Research - Local to Global)

• Nicole Pratt (University of South Australia)
• Graeme Hart (University of Melbourne)
• Clair Sullivan (Queensland Health)
• Roger Ward (ARDC)

**Significance of the Issues** Australia’s healthcare system generates a vast amount of data, however, data systems are highly fragmented, with information captured across diverse and often incompatible systems. This lack of interoperability creates major barriers to the integration and analysis of health data at scale, limiting the nation’s ability to conduct efficient, multi-centre research and generate timely, actionable evidence for health policy and clinical care. Internationally, federated data networks such as the European Health Data and Evidence Network (EHDEN) and the Observational Health Data Sciences and Informatics (OHDSI) community have demonstrated the value of a harmonised infrastructure using the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM). These models promote standardisation, preserve privacy, and enable research that is scalable, reproducible, and globally collaborative. To address this critical need in the Australian context, The University of South Australia (UniSA), with co-investment from the Australian Research Data Commons (ARDC), has established the Australian Health Data Evidence Network (AHDEN). AHDEN's mission is to build a nationally coordinated infrastructure that supports the transformation of hospital-based Electronic Medical Record (EMR) data into the OMOP CDM format across Australia. The OMOP CDM is and open-source, internationally recognised data standard that enables consistent structuring of observational health data across multiple different sources such as hospitals, primary care systems, and disease registries. By transforming diverse local healthcare data into the OMOP CDM format, AHDEN will enable researchers to apply a common suite of analytic tools and methods to gain insights more efficiently, enhancing reproducibility and supporting collaborative research without compromising data security or privacy. Ultimately, AHDEN will strengthen national research capacity, foster data-driven health policy, and unlock the full potential of real-world data for improving population health. **Approach, Structure, and Format** This Session will showcase the AHDEN initiative and the power of the OMOP CDM to enable a scalable, federated infrastructure for health data research. Presentations will demonstrate the technical foundations of the OMOP CDM format, including syntactic and semantic harmonisation, as well as approaches to governance, security, and distributed analytics. The session will highlight the strengths of the OMOP CDM in enabling privacy-preserving data analytics at local, national, and international scales. Participants will gain practical insights into the challenges and successes of mapping Australian EMR data to the OMOP CDM and see how this infrastructure supports the generation of evidence for regulatory, clinical, and policy applications. **Proposed Speakers and Topics** 1) Professor Nicole Pratt, AHDEN Project Lead, University of South Australia. (10 minutes) Professor Pratt will introduce the vision and mission of AHDEN and outline the national and international significance of building a federated health data infrastructure. 2) Associate Professor Graeme Hart, University of Melbourne (15 minutes) Associate Professor Hart will describe the technical processes involved in syntactic and semantic harmonisation of health data using the OMOP CDM, highlighting the challenges and solutions in applying this model to Australian EMR systems 3) Roger Ward, Solutions Architect, Australian Research Data Commons (15 minutes) Mr Ward will focus on the data governance and privacy-preserving infrastructure that underpins AHDEN. He will explain how the federated model ensures compliance with privacy regulations while enabling scalable research. 4) Professor Clair Sullivan, University of Queensland (15 minutes) Professor Sullivan will present progress on mapping Queensland’s statewide EMR observation data to the OMOP CDM. She will highlight the clinical and operational benefits of standardisation at a jurisdictional level. 5) Professor Nicole Pratt, AHDEN Project Lead, University of South Australia) (15 minutes) Professor Pratt will present a series of impactful clinical applications that have been enabled through the use of the OMOP CDM, including evidence generation of the safety and effectiveness of all second-line diabetes medications, real-time response to emerging health threats during the COVID-19 pandemic, and investigation of rare adverse events supporting global pharmacovigilance efforts for medicine regulators. 6) Panel discussion (20 minutes) The session will conclude with a 20-minute panel discussion where all speakers will respond to audience questions and reflect on the future of federated research in Australia This session directly addresses the growing demand for national-scale health data infrastructure that is privacy-preserving, methodologically rigorous, and interoperable. By showcasing AHDEN’s approach to standardising EMR data through the OMOP CDM, the session will demonstrate the feasibility and value of federated health data research in Australia. Attendees will leave with a clear understanding of how this infrastructure supports timely, high-quality evidence generation that can inform clinical practice, regulatory decision-making, and health policy both nationally and globally. **Contribution Type:** Session

AI without borders? Navigating data sovereignty and human rights in a fragmented world (Empowering the global data community for impact, equity, and inclusion)

• Alexander Kriebitz (Technical University of Munich (TUM))
• Caitlin C. Corrigan (Technical University of Munich (TUM))
• Perihan Elif Ekmekci
• Francis P. Crawley (CODATA IDPC)

Introduction This workshop will examine how the evolving architectures of data policy and AI governance challenge traditional notions of political sovereignty, including state sovereignty, the sovereignty of the people, and the sovereignty of the individual human being, as recognized in the United Nations Charter and the Universal Declaration of Human Rights (UDHR). The United Nations Charter upholds the “sovereign equality of all its Members” (Article 2.1), affirming the right of states to self-determination and governance without external domination. However, the dominance of transnational tech corporations, extraterritorial data flows, and private AI infrastructures now blur jurisdictional boundaries and pose direct challenges to state control over digital infrastructures and decision-making processes. This has led to a growing demand for digital and data sovereignty—where states assert the right to regulate, localize, and govern data produced within their borders. The UDHR, in parallel, affirms the inherent dignity, freedom, and equality of all individuals (Articles 1 and 2) and protects individual sovereignty through rights such as privacy (Article 12), freedom of opinion and expression (Article 19), and participation in governance (Article 21). Yet, automated systems that profile, manipulate, or surveil citizens challenge these very freedoms. The session will draw on practical and policy insights from global processes such as: • The United Nations Charter • The Universal Declaration of Human Rights (UDHR) • The Munich Convention on AI, Data, and Human Rights (2024–25) • The whitepaper: ‘Promoting and Advancing Human Rights in Global AI Ecosystems: The Need for A Comprehensive Framework under International Law’ • The African Commission’s Draft Study on Human and Peoples’ Rights and AI (2025) • The UN Global Digital Compact (2024) • The UNESCO’s Recommendation on the Ethics of Artificial Intelligence (2021). Format Workshop Format and Detailed Agenda Interactive Workshop Session – 90 minutes This session is designed to foster deep engagement, practical insight-sharing, and collaborative policy thinking across global, regional, and disciplinary divides. It combines expert framing with structured group discussions and plenary synthesis to address complex questions around sovereignty, AI, and human rights. The session format follows four core principles: 1. Participatory: All attendees contribute to dialogue, not just listen. 2. Multi-perspective: Structured to reflect legal, ethical, cultural, and geopolitical diversity. 3. Action-oriented: Designed to generate recommendations and pathways for future work. 4. Grounded: Built on real frameworks like the UN Charter, UDHR, Munich Convention, and regional AI strategies. Workshop Agenda (90 Minutes) Additional speakers tbc: • African Commission on Human and Peoples’ Rights representative (TBC) – Regional perspectives on collective and peoples’ rights in AI governance • Global South civil society or technologist (TBC) – Challenges of asserting local digital sovereignty in the face of global AI platforms 1. Opening and Scene-Setting Presentations (20 minutes) Moderated by session chair Purpose: Establish foundational concepts of sovereignty (state, peoples, individual) and introduce global legal and policy frameworks relating to AI and human rights. • 5 min — Introduction and workshop goals • 5 min — Presentation: "Sovereignty in the Age of AI: From the UN Charter to the Digital Realm" • 5 min — Regional insight: African and Latin American perspectives on sovereignty and AI • 5 min — Legal perspective: The Munich Convention on AI, Data, and Human Rights Speakers: Global experts from law, ethics, and policy communities, including contributors to the Munich Convention. 2. Breakout Group Discussions (30 minutes) Participants will divide into four facilitated breakout groups, each tackling a key sovereignty dimension. Each group includes a facilitator and rapporteur and will address 2–3 guiding questions rooted in current international law and AI ethics frameworks. Group 1 – State Sovereignty Topic: National governance of cross-border AI and data regimes Focus: Data localization, extraterritorial digital infrastructure, compliance with human rights Group 2 – Sovereignty of Peoples Topic: Collective rights, digital democracy, and algorithmic governance Focus: Citizen participation, access to information, community data rights Group 3 – Individual Sovereignty Topic: Autonomy, data protection, and consent in AI systems Focus: Privacy (UDHR Article 12), meaningful consent, surveillance ethics Group 4 – Bridging Sovereignties Topic: Harmonizing local values with global digital governance Focus: Cross-cultural rights recognition, UN principles vs. local ethics, legal interoperability Participants will discuss: • Opportunities to use AI and sovereignty to advance human rights • Key risks and legal/policy gaps • Practical actions for future governance 3. Plenary Synthesis and Panel Response (30 minutes) Each group reports back (5 min per group), followed by discussion. • 20 min — Plenary synthesis: Group rapporteurs present key findings • 10 min — Panel response: Short reflections from a multi-region expert panel (Africa, Asia-Pacific, Latin America, Europe) This segment will explore cross-cutting tensions and possible reconciliations between sovereignty and universal rights in AI governance. 4. Closing and Outcome Framing (10 minutes) Moderator summarizes key outcomes and presents a draft outline for a post-workshop document: “Action Note on Sovereignty and Rights in AI Governance” • Invitation for ongoing collaboration via CODATA, CoARA-ERIP, or a dedicated follow-up task force • Next steps for aligning this work with SciDataCon and IDW open science values Participants will be invited to co-sign or contribute to the post-workshop output. Expected Outcomes • A deeper understanding of how international human rights frameworks interact with emerging digital sovereignties • Strategic insights into how sovereignty claims can be balanced in AI governance across state, community, and individual levels • Contribution to the formation of a global action framework or policy brief on AI, sovereignty, and rights-based digital governance

AI for Metadata Enhancement, Metadata for AI Readiness: how do we ensure a virtuous rather than a vicious circle? (Rigorous, responsible and reproducible science in the era of FAIR data and AI)

• Doug Fils (Consultant / CODATA)
• Rebecca Farrington (AuScope)
• Sean Hill (SenScienceAI)
• Simon Hodson (CODATA)
• Jieping Ye (GeoGPT / Zhejiang Lab)
• Pascal Heus (Postman / CODATA)
• Cristina Gonzalez (SenScienceAI)
• Deirdre Lungley (UKDS)
• Vyacheslav Tikhonov (DANS)

This session will explore the intriguing and potentially urgent interaction (and even codependency) between high quality metadata and semantic richness on the one hand and Generative Artificial Intelligence (AI) and Large Language Models (LLMs) on the other. A lot of work is going on to improve the richness, quality and standardisation of metadata and semantics in order to make data sets ‘AI ready’. At the same time, the potential of generative AI is being explored precisely to enrich metadata and semantics. Exercising caution in this endeavour is critical, however, as the quality of the outputs is directly tied to the quality of the underlying data and documentation. The topic of this session is to explore, through numerous examples and discussion, the latest work in this area. AI for Metadata Enhancement, Metadata for AI Readiness: how do we ensure a virtuous rather than a vicious circle? On the side of metadata for AI readiness, we have: - [ML Commons’ Croissant][1] - The [Cross-Domain Interoperability Framework (CDIF)][2] is aligning with Croissant and has important components on data description (a profile of [DDI-CDI][3]) and work underway on provenance and data quality, which is important in this context. - Work on semantic mappings and knowledge graphs. On the side of AI for metadata enhancement and inference we have: - The remarkable work of the SenScience team with [FAIR2][4] and a compelling example of metadata enhancement for a [Frontiers data article and science article on biodiversity off the Basque coast][5]. - The work at [Closer][6], UK on metadata inference, enrichment and [‘uplift’][7]. - [GeoGPT][8] assisted classification and application of geological terminologies and semantics. In parallel, there is a growing realisation that maintaining the quality of AI metadata enhancement and inference, requires the LLMs being able to access key knowledge, for example through a Model Context Protocol server, as expressed in authoritative terminologies or other sources of reference: - The idea of a Model Context Protocol (MCP) server for the [SI Reference Point][9] to make the underlying knowledge accessible to LLMs and agent. - The idea of implementing a MCP server for Croissant and for CDIF. - Work to predict semantic mappings (including GeoGPT and the work of Vyacheslav Tykhonov). The session will be composed of a number of quick-fire presentations covering various aspects of the issues raised here, thus below, in many instances, we give not titles but issues and examples to be introduced and discussed. We intend this to be a rapid exchange of ideas rather than a series of formal presentations. There will be significant time for discussion. One outcome will be a quick report surveying the landscape and covering the issues raised. Above all, however, we will seek to identify concrete steps that scientific communities and the Research Infrastructures that serve them can take, drawing on these examples and emerging practices, to address issues of AI readiness and metadata enhancement, while ensuring we achieve a virtuous circle. **Programme:** Simon Hodson, Arofan Gregory, CODATA: - Introduction to the issues: AI for Metadata Enhancement, Metadata for AI Readiness: how do we ensure a virtuous rather than a vicious circle? Doug Fils, Consultant / CODATA; Vyacheslav Tykhonov, DANS; Pascal Heus, CODATA / Postman: - Croissant, Semantic Croissant and GeoCroissant. Pascal Heus, CODATA / Postman: - The critical role of FAIR Open Data APIs for AI - Findings of the CDIF AI Readiness Working Group - Related R&D and topics Vyacheslav Tykhonov, DANS: - AI-powered Semantic Mappings with RAG for ontology alignment - Leveraging AI to Automatically Link Controlled Vocabulary Terms in Metadata - Semantic Croissant: Enabling FAIR Data for AI Applications with the Model Context Protocol (MCP) - MCP Server Library: A Foundation for AI Applications and FAIR Data Workflows - Dataverse: Building a Distributed Data Network Ready for AI Deirdre Lungley, UKDS: - AI for Metadata Enhancement and Inference: the example of UKDS metadata uplift. - Metacurate-ML: UK ESRC funded project to improve curation tooling, enabling semi-automated metadata uplift at scale. Workstreams: - Questionnaire Extraction from PDFs using LLMs - Subsequent Question alignment using AI - LLM topic classification of these questions/variables - Harnessing the knowledge graph produced in these preceding steps, together with further LLM identification of indirect identifier variables to power dataset ingest, including Statistical Disclosure Control (SDC) Sean Hill, Cristina Gonzales, SenScience: - FAIR2 Jieping Ye, Zhejiang Lab / GeoGPT: - GeoGPT for classification. Rebecca Farrington, AuScope; Kelsey Druken, ANU; Isabel Ceron, Australian Academy of Social Sciences: - AI readiness and metadata inference in Australia and beyond **Discussion:** - Landscape - Future collaborations - Recommendations [1]: https://mlcommons.org/working-groups/data/croissant/ [2]: https://cross-domain-interoperability-framework.github.io/cdifbook/introduction.html [3]: https://ddialliance.org/ddi-cdi [4]: https://www.frontiersin.org/about/fair-data-management [5]: https://www.frontiersin.org/journals/ocean-sustainability/articles/10.3389/focsu.2024.1528837/full [6]: https://closer.ac.uk/ [7]: https://ukdataservice.ac.uk/about/research-and-development/past-projects/cohort-and-longitudinal-studies-enhancement-resources-closer/ [8]: https://geogpt.zero2x.org/ [9]: https://www.si-digital-framework.org/

An evolving role for Data Scientists in the Age of Intelligent Automation (Data Science and Data Analysis)

• Mark Parsons
• Matthew Mayernik (NSF National Center for Atmospheric Research)
• Dimitris Symeonidis (University of Tartu)
• Gita Yadav (National Institute of Plant Genome Research (NIPGR))
• Debasis Mohanty (National Institute of Immunology, New Delhi)
• LILI ZHANG (COMPUTER NETWORK INFORMATION CENTER, CAS)

Data science has evolved significantly over the past two decades, becoming a force within academic research, public and private sector workplaces, and in government policies and practices. Exponentially increasing volumes of publicly available datasets throughout the social and scientific realms have contributed to an explosion of data science applications, including AI tools and Large Language Models (LLMs) that are being connected to digital technologies of all types, such as personal computers, cell phones, social media, smart devices, and sensor networks. Data science has become a broad term with very different meanings and instantiations, growing far beyond the traditional panoply of techniques applied to derive value (economical, intellectual, or cultural) out of data. A former editor of the CODATA Data Science Journal argued in a recent retrospective essay that, “The scientific data community requires more from data science than other communities” (Rumble, 2023, p. 2), arguing that topics like data preservation, provenance, and traceability cannot be ignored when developing data science applications for scientific research. In this session, we will present broad perspectives on this key question: What are the distinctive aspects of data science for the scholarly and scientific data community, and how should the present day data scientists adapt to address issues like FAIRification, AI-readiness and Ethics? A diverse panel will engage the CODATA community in a discussion of the role of data science in relation to scientific data initiatives, and how this community is contributing to the growth and evolution of data science. The speakers in this session will address several important questions about the current status and future evolution of data science, including: - What is data science, and how does it relate to the “science of data”? - Is data science a scientific domain in its own right? - What are the main gaps / opportunities that must be addressed in data science going forward? - What are the key data science trends in relation to the scientific data community? - Science has always been data driven; what is different now? How have changes in the way that data are or should be shared influenced data science? - How do AI technologies impact data science and contribute to FAIR research data ecosystems? - How should the Data Science Journal and SciDataCon respond to changes in the nature of data science? The talks in this session will inform broader discussions of the implications of these questions for CODATA, the World Data System (WDS), the International Science Council, and the many international scientific associations and unions. How can these organizations both contribute and respond to the changing nature of data science via concrete actions, programmes, and initiatives? References: Rumble, J. (2023). Thoughts on starting the CODATA Data Science Journal. Data Science Journal, 22, 13. https://doi.org/10.5334/dsj-2023-013 Speakers: - Matt Mayernik (Session co-Moderator) - NSF National Center for Atmospheric Research, USA, Editor-in-Chief, Data Science Journal - Gita Yadav (Session co-Moderator) - Scientist, National Institute of Plant Genome Research (NIPGR), New Delhi India; Member, CODATA IDPC; Professor of Data Science, IISER Bhopal, India Founder, #SemanticClimate; Member Editorial Team, Data Science Journal - Deb Mohanty - Director, National Institute of Immunology, DBT, Govt of India; Member, CODATA India National Committee; Chairperson, Indian Biologica Data Center - Mark Parsons - Former Editor-in-Chief, Data Science Journal; Member, Arctic Data Committee for the International Arctic Science Committee (IASC) and the Sustaining Arctic Observing Networks (SAON) CODATA representative to the International Polar Year 2032-3 Planning Committee - Dimitris Symeonidis - University of Tartu, Institute of Computer Science Member, Digital Government Society - Dr. Lili Zhang - Executive Director, GOSC IPO, member of CODATA IDPC Senior Research Scientist, CSTCloud, CNIC,CAS; Member Editorial Team, Data Science Journal

"So much going on!" How to best coordinate international efforts for data management — a polar to global case study and discussion (Data Stewardship)

• Chantelle Verhey
• Michaela Miller
• Mark Parsons

As data managers and practitioners, we've all felt it: "I want the data I handle to be broadly discoverable,interoperable, and preserved according to described best practice, but I first have to address the demands of my data providers and users who have their own special (and well-loved) way of doing things. I want to make sure I'm plugged into the leading practices of the global data management community while also making sure my domain/community is represented and heard. I need to solve my local problems in a way that fits into a global data ecosystem. I need to be in multiple places and spaces at once!" Meanwhile, we are living in the world of big data and open science. The Fourth Paradigm is here. The possibilities for machines and algorithms to process and help interpret massive amounts of diverse data continues to accelerate while the array of scientific and operational initiatives collecting that data continues to grow. All this is testing human organizational capacity. The polar science community has a long history of international collaboration and has been coordinating data since the first International Polar Year in 1882. Major data coordination initiatives continue today including the recently concluded MOSAiC (Multidisciplinary drifting Observatory for the Study of Arctic Climate) expedition, the Antarctica InSync (Antarctica International Science & Infrastructure for Synchronous Observation) initiative, as well as the next IPY in 2032 (IPY5). Correspondingly, the polar data community seeks to learn from the experiences of the global community as well as other regions and disciplines. Using IPY5 as a milestone, the polar data community seeks to foster more effective and more efficient data collaboration at all levels. In this session, we will introduce current large-scale data coordination activities and challenges. We will then conduct a structured discussion around “integrated autonomy” to begin to identify “How is it that we can be more integrated and more autonomous at the same time?” The goal is to identify several focus areas for international collaborations (i.e. standardization) that also address local needs (i.e customization). **Session Speakers and Structure:** Introduction and Context (10 min) Mark Parsons CODATA Polar Data Advocate - Overview of major data coordination activities and challenges in the Antarctic (10 min) Michaela Miller Data Officer, Southern Ocean Observing System - Overview of major data coordination activities and challenges in the Arctic (10 min) Chantelle Verhey Co-chair, Arctic Data Committee - Structured Discussion (55 min) -- Introduction to the exercise “How is it that we can be more integrated and more autonomous at the same time when addressing the needs of large-scale science initiatives?” (5 min) -- Small groups work to: --- Identify tensions between a desire to standardize and the request for more customizing or autonomy?” (10 min) --- Pick an activity from the list above and ask “What is the rationale for standardizing? What is the rationale for customizing?” Then develop action steps that achieve standardization and action steps that achieve customization. (10 min) --- Identify actions that boost both standardization and customization and what modifications can be made to some actions so that they boost both standardization and customization. (10 min) --- Prioritize the most promising actions that promote both integration and autonomy. (10 min.) -- Report out (10 min) -- Closing and follow up (5 min)