D5.5 - Publication on the blueprint of the national and the pan-European STRONG AYA ecosystems

1 A new, interdisciplinary, multistakeholder European network to improve healthcare services, research and outcomes for Adolescents and Young Adults with cancer. STRONG-AYA – No. 101057482 – D5.5 Deliverable Report WP5– Scientific coordination and project management Deliverable D5.5 Publication on the blueprint of the national and the pan-European STRONG AYA ecosystems Due date of deliverable: 30/09/2024 Actual submission date: 11/09/2025 Project: STRONG-AYA Lead Contributor Dan Stark, Leonard Wee Email [email protected]; [email protected] Other Contributors Oana Lindner, Simone Hanebaum, Nicola Hughes, Urska Kosir, Winette Van Der Graaf, Olga Husson Emails [email protected]; [email protected]; [email protected]; [email protected]; [email protected]; o.hus[email protected] Due date 30 09 2024 Delivery date 11 09 2025 Deliverable type R Dissemination level PU Description of Work Version Date V1.0 11/09/2025 STRONG-AYA – No. 101057482 – D5.5 Description: Scientific paper Publishable summary (max ½ page) The STRONG-AYA consortium is addressing the challenge of using real world data to understand the clinical and patient-centred outcomes that are specific to cancer diagnosed aged 15-39. We will use this data to improve care, policy, and ultimately the outcomes themselves. A harmonised 'Core Outcome Set’ is agreed, to define what should be measured. We are building a series of AYA cancer data ecosystems, at local, national and international levels, that will use federated learning methods to provide pre-specified analyses in real time, based upon fundamental ethical, scientific and technological principles. Here we describe the problem being addressed. Then, for others interested in AYA cancer to understand and join our initiative, we describe: - The definitions and decisions we have made. - The structure and components of our ecosystem - The context of our ecosystem alongside similar others, and our approaches to challenges we face STRONG-AYA – No. 101057482 – D5.5 1 Table of Contents PUBLISHABLE SUMMARY (MAX ½ PAGE) .................................................................................................................. 3 1 TABLE OF CONTENTS ....................................................................................................................................... 4 2 DEFINITIONS .................................................................................................................................................... 6 3 ABBREVIATIONS .............................................................................................................................................. 7 4 DELIVERABLE INTRODUCTION/CONTEXT .......................................................................................................... 8 DELIVERABLE INTRODUCTION .................................................................................................................................... 8 5 INTRODUCTION ............................................................................................................................................... 8 AYAS SPECIFIC CANCER CARE NEEDS ........................................................................................................................... 8 THE REQUIREMENT FOR AYA-SPECIFIC CANCER OUTCOMES DATA .................................................................................... 9 THE STRONG AYA INITIATIVE ............................................................................................................................... 10 DEFINITION OF A DATA ECOSYSTEM IN CANCER CARE .................................................................................................... 10 6 DATA ECOSYSTEMS WITHIN STRONG-AYA (TABLE 1) ...................................................................................... 11 LOCAL STRUCTURES .............................................................................................................................................. 11 PARTLY NATIONAL STRUCTURES ............................................................................................................................... 11 NATIONAL STRUCTURES ......................................................................................................................................... 12 INTERNATIONAL STRUCTURES .................................................................................................................................. 12 7 GOVERNANCE MODEL, FOR THE STRONG-AYA DATA ECOSYSTEM .................................................................. 15 THE FOUNDING INTERPERSONAL AND ETHICAL PRINCIPLES OF THE STRONG-AYA DATA ECOSYSTEM .................................... 15 7.1.1 Ethical, value-based healthcare .................................................................................................................. 16 7.1.2 Inclusive and focused on patient benefit .................................................................................................... 16 7.1.3 Open and transparent science .................................................................................................................... 17 GOVERNANCE IN STRONG-AYA ............................................................................................................................ 18 7.2.1 Definition of federated learning .................................................................................................................. 19 7.2.2 Legal governance supporting federated data ecosystems ......................................................................... 20 PERSONNEL IN STRONG-AYA ............................................................................................................................... 21 DATA ANALYTICS TECHNOLOGY AND TECHNICAL STANDARDS IN STRONG-AYA ................................................................ 21 7.4.1 Autonomy of local and national data collection structures ........................................................................ 22 7.4.2 Technical implementation of data ‘stations’ .............................................................................................. 22 7.4.3 Statistical computing code as standalone ‘trains’ ...................................................................................... 23 7.4.4 Data analytics dashboards .......................................................................................................................... 23 7.4.5 Privacy enhancements in STRONG-AYA ...................................................................................................... 24 7.4.6 Challenges encountered in the federated analytics approach .................................................................... 24 8 WHAT CAN WE LEARN FROM SIMILAR INITIATIVES? ...................................................................................... 25 9 KEY CHALLENGES FACING THE STRONG AYA CONSORTIUM AND OTHER DATA ECOSYSTEMS .......................... 26 SCALABLE, SUSTAINABLE, AND INTEROPERABLE ........................................................................................................... 26 MANAGEMENT OF NATIONAL AND INTERNATIONAL DIFFERENCES, OVER TIME ................................................................... 26 MANAGEMENT OF WIDER STAKEHOLDER ACCESS – WITHIN AND OUTSIDE OF THE CONSORTIUM ........................................... 28 ENGAGEMENT AND ‘BUY IN’ ................................................................................................................................... 29 STRONG-AYA – No. 101057482 – D5.5 9.4.1 Technological buy-in .................................................................................................................... 29 9.4.2 Data hesitancy (patients, governments, IP and academic politics, registry redundancy fears, etc) ........... 30 RESOURCES (DIGITAL, FINANCIAL, PERSONNEL, EXPERTISE ETC.) ..................................................................................... 31 LONGEVITY/SUSTAINABILITY ................................................................................................................................... 32 10 FUTURE ACTIONS AND CONSIDERATIONS ...................................................................................................... 32 11 REFERENCES .................................................................................................................................................. 33 STRONG-AYA – No. 101057482 – D5.5 2 Definitions STRONG AYA consortium members are referred to as following within this text: 1. NKI-AVL – Stichting het Nederlands Kanker Instituut – Antoni van Leeuwenhoek Ziekenhuis (NL) 2. YCE – Youth Cancer Europe (RO) 3. INT – Fondazione IRCCS Instituto Nazionale dei Tumori (IT) 4. FFUND – FFUND BV (NL) 5. CLB – Centre de Lutte Contre le Cancer Leon Berard (FR) 6. ECO – European Cancer Organisation (BE) 7. UNIMAAS – Universiteit Maastricht (NL) 8. IKNL – Stichting Integraal Kankercentrum Nederland (NL) 9. EORTC – European Organisation for Research and Treatment of Cancer AISBL (BE) 10. IGR – Institut Gustave Roussy (FR) 11. MSCNRIO – Narodowy Instytut Onkologii im. Marii Sklodowskiej-Curie – Panstwowy Instytut Badawczy (Marie Sklodowska-Curie National Research Institute of Oncology) (PL) 12. UOM – University of Manchester (UK) 13. UOL – University of Leeds (UK) 14. LTHT – Leeds Teaching Hospitals National Health Service Trust (UL) 15. SOUTHAMPTON – University of Southampton (UK) • Grant Agreement (including its annexes and amendments): the agreement signed between the beneficiaries of the HORIZON Research and Innovations Actions (hereafter referred to as Horizon) and the European Health and Digital Executive Agency (hereafter referred to as HADEA) for the undertaking of the STRONG AYA project (Grant Agreement no. 101057482). • Beneficiary: Signatories of the Grant Agreement • Associated Partner: Entities which participate in the action but without the right to charge costs or claim contributions. • Project: the sum of all activities carried out in the framework of the Grant Agreement. • Consortium: the STRONG AYA consortium, including all the aforementioned partners. • Consortium Agreement: The agreement made between STRONG AYA members for the implementation and execution of the action outlined in the Grant Agreement. The agreement shall not affect the parties’ obligations to HADEA on behalf of the European Union, and/or to one another arising from the Grant Agreement. STRONG-AYA – No. 101057482 – D5.5 3 Abbreviations Acronym/Abbreviation Meaning HCP Health Care Provider PRO Patient Reported Outcome PROM Patient Reported Outcome Measure COS Core Outcome Set WP Work Package WPL Work Package Lead(s) WP1 Work Package 1 (Development Core Outcome Set AYA with cancer & data collection) WP2 Work Package 2 (Governance, Data Security and Ethics) WP3 Work Package 3 (Infrastructure and Interoperability) WP4 Work Package 4 (Operation of STRONG AYA ecosystems, stakeholder and patient involvement, dissemination, exploitation, communication) WP5 Work Package 5 (Scienti fic coordination and project management) KPI Key Performance Indicator OA Open Access PAB Patient Advisory Board EC European Commission HADEA European Health and Digital Executive Agency SC Steering Committee MT Management Team STRONG-AYA – No. 101057482 – D5.5 4 Deliverable Introduction/context Deliverable introduction The following deliverable is the manuscript for the ecosystem blueprint for STRONG AYA. It will be submitted for publication in a shorter length but we wished to preserve its detail and full length via this deliverable. 5 Introduction International healthcare systems aim to provide high-quality care and improve patient outcomes, ranging from clinical parameters to symptoms and optimized quality of life. Achieving these goals relies heavily on the accurate recording, monitoring, and evaluation of healthcare quality indicators (such as clinical effectiveness, patient safety, patient-centred and patient-reported outcomes - 'PROs'). This data increasingly arises from real-world data (RWD) collection and is then utilised within ‘learning healthcare systems’ as the evidence base for care transformation. Cancer in Adolescents and Young Adults (AYAs) are either rare cancer types, such as sarcomas, lymphomas, and germ cell malignancies, or uncommon presentations of common cancers e.g. early-onset carcinomas. Therefore, generating practice-changing evidence supported by high quality data is particularly challenging. This challenge is exacerbated because indicators of care quality and PROs have been (up to now) collected sporadically, or non-systematically, and the most important outcomes to measure are lacking a consensus. This makes it harder to prioritize and implement the right healthcare improvements for the AYA cancer population from a strong evidence base. AYAs specific cancer care needs In our scope, we selected the existing European gold standard, defining AYAs with cancer as people receiving a cancer diagnosis between 15 and 39 years of age inclusive. First (for the individuals with cancer) there are the challenges of AYA receiving optimal cancer diagnosis, treatment and care. Worldwide, an estimated 1.2 million AYAs 1 were diagnosed with cancer in 2020, representing around 5% of all cancer diagnoses 2. The incidence of AYA cancer is increasing over time 3. AYA-specific diagnosis and treatment has recently improved, since this patient group became a focus of specific focus and research. All-cause and cancerrelated mortality has fallen in high-income countries 4, so there over 85% of AYAs with cancer will survive ≥5 years 5,6 and some will live >50 years beyond their treatment7. If they receive the most effective cancer care (which should be theirs by right) then the individual AYAs with cancer join a rapidly growing cohort of cancer survivors who will spend the remainder of their lives with increased risks of physical, psychological, and social effects; cardiovascular disease, secondary malignancies, infertility, financial toxicity, and premature mortality 8-10. These late effects are also long recognised as specific to the AYA age range 11. Until recently, across many countries, healthcare professionals (HCP), policy-makers and other stakeholders possessed limited knowledge about this population's specific needs 12. Sadly, cancer in AYAs is often diagnosed, treated and cared for by either paediatric (<15-18 years) or adult (>16) care specialists, when in fact AYAs are liminal for both those services (caring mainly for patients aged under 12 or over 70 respectively) and are a distinct group, with a specific set of priority challenges in clinical practice, requiring a specific focus to address 1, 2, 13-15. AYAs’ challenges pertain to their: STRONG-AYA – No. 101057482 – D5.5 (1) Unique epidemiology - AYAs develop both paediatricand adult-type tumours, requiring expertise from both clinical services 16. (2) Distinct tumour biology - cancers in AYAs often exhibit age-specific therapeutically significant molecular features and pharmacological responses, compared to younger children or older adults with superficially similar situations16. (3) Delayed diagnosis - a lack of awareness about cancer in AYAs and silos in services contribute to diagnostic delays and therefore poorer outcomes 17,18. (4) Limited access to clinical trials - AYAs participate in clinical trials at lower rates due to unavailability of tailored treatments and trial designs that do not meet their specific needs, resulting in slow improvement in outcomes 19. (5) Personal Development - AYAs are undergoing life transitions (e.g. identity formation, autonomy, and career development) which intersect with their cancer, making both healthy development and best cancer treatment more difficult to achieve 13,20. (6) Survival rate disparities - survival improvements for AYAs lagged behind those seen in other age groups due to the factors above, unless systemic gaps in care and research are addressed 21. The requirement for AYA-specific cancer outcomes data In services where they are liminal, AYAs with cancer struggle to find care that manages their cancer in keeping with their age 22. Their unique support needs are often not adequately met by centralised paediatric services (and “family-centred” teams) nor dispersed adult oncology services (and “disease-centred” teams) 14,23,24. Recently, AYA cancer services in some parts of the world have been accepted as a specific distinct part of a complete cancer service, and provided with AYA-specific cancer care14. Maintaining this recent momentum requires ongoing research improvements, to define the data required to understand this progress, and to continue to improve outcomes. In 2016, the National Cancer Institute (NCI) AYA cancer progress review group report requested AYA caregivers to pool process-related and outcomes data across institutions and countries, to make large cohorts available for research 12,22. Extending these recommendations, the AYA Working Group of the European Society for Medical Oncology (ESMO) and the European Society for Paediatric Oncology (SIOP Europe) concluded that finding rapid solutions to ‘speak the same language’ among AYA cancer professionals is essential to further improve outcomes for AYAs. Meanwhile widespread geographic variation in AYA cancer policy and care persist within and between nations, and equitable services and optimal care for AYAs have not yet been achieved 14. Collection of all cancer outcomes data is currently not standardized, and is implemented unevenly across health services; some European health services routinely record and monitor clinical outcomes as well as PROs, but many do not. Commonly-used clinical trial endpoints (such as five-year overall survival) are important, but these do not reflect AYAs’ specific needs, and neither do outcome measures designed for young children, older adults, or any site-specific cancer type. Where data is collected (with more and better data to be collected in future) the task of “pooling” highly granular patient data into large repositories runs into challenges due to differences in research governance structures, rules for data re-use, nuances and interpretations of patient privacy protection laws. Therefore, novel approaches to perform large-scale statistical analyses across multiple dispersed locations are needed, which can act in a complementary fashion to traditional data centralization into singular repositories. STRONG-AYA – No. 101057482 – D5.5 • Data curator (quality and availability) Figure 2: a broad taxonomy of potential roles within a data ecosystem, adapted from S. Oliveira, M.I., Barros Lima, et al., 2019. The alignment of incentives and motives across these roles relies on a keen understanding of each individual and institution’s capacity to achieve their goals through the joining of a pan-European data ecosystem. The ongoing incentives and motives for the existence of the STRONG-AYA Consortium and pan-European ecosystem are reflected in the principles outlined in more detail further, but are here summarized as: 1. Improvement in the provision of healthcare services: a. The development of value-based clinical practice; b. Accelerating the implementation of real-world insights to improve clinical practice. 2. Performing ethical research on large datasets: a. Encouraging international collaboration for research and clinical expertise; b. Improving and expanding upon the applications of available data in research; c. Understanding and leveraging institutional capacity. 3. Improving patient outcomes: a. Defining and identifying patient unmet needs – in research and clinical fields; b. Improving the diversity and reach of participants and beneficiaries from research and patient outcome measurement; c. Improving the utilisation of real-world health and healthcare data. We believe it perfectly reasonable that different members may have a range of incentives for participating in the Consortium. Institutional incentives and motivations might also go beyond direct patient benefit, which naturally influences the purposes of data analysis that they conduct. These may encompass discovery (i.e. mining increasing volumes of data to help manage rare diseases), or accelerating implementation of existing initiatives (i.e. sharing administrative workload on diverse datasets), or encouraging sustainable international collaboration (i.e. tangible outcomes that foster collaborative practices). Below we provide more detail on the foundational principles of STRONG-AYA, which drive the structure and functioning of the components as defined above. 7.1.1 Ethical, value-based healthcare By broadening the research using novel privacy-safe approaches towards data analytics, STRONG-AYA will bring novel insights into AYA healthcare, encourage ethical re-use of scarce AYA research data, provide realworld evidence to all important AYA oncology decision-makers, and thus, increase the benefit for patients by making care more efficient. 7.1.2 Inclusive and focused on patient benefit STRONG-AYA tools and activities are patient-centred, following the ethos of ‘nothing about me without me’. These ethical principles, centred on the interests, preferences, and values of patients, are foundational to the Consortium, reiterated and reflected in the design, built and maintained through stakeholder STRONG-AYA – No. 101057482 – D5.5 engagement and reflections by the Patient Advisory Board (PAB). The PAB is composed of young people with lived experience, composed based on Equity Diversity and Inclusion principles, actively seeking representation across various dimensions such as expertise, geography, and tumour type. The PAB provides ongoing support for the STRONG-AYA initiative by sharing knowledge, facilitating feedback with AYA cancer patient communities, and contributing to outreach, webinars, and other collaborative activities. Importantly, any tools and actions within STRONG-AYA aim to complement and optimize face-to-face healthcare; patients are informed of the benefits and the limits of tools developed by the Consortium, and can customize their interactions with these tools. STRONG-AYA's resulting insights and tools need to be accessible, meaning patients: • can easily and reliably access/retrieve their own STRONG-AYA health data (usually through their own hospitals’ systems); • see and better understand their own health data in the context of other, similar, AYAs; • easily access information on the uses of their health data, and its’ purpose, such that they should be able to easily enhance, grant or remove access to their health data depending upon their perspective on each ‘case of use’ of their data, and exercise this right freely; • will have a voice in designing tools developed by STRONG-AYA, including promoting the principles of EDI such that the solutions remain accessible to all (including for example people with disabilities, neurodivergent populations, or different levels of digital literacy). They should be intuitive and easy to use; patients should have access to training to help them understand these tools, which should be implemented in routine care that includes human communication, space and time for additional questions and feedback. Hence, STRONG-AYA will not only equip AYAs with lived experience with information to optimize their healthcare, but also offer an opportunity to build on their health literacy, and support shared and personalized decision-making between AYA and health care providers. This extends to involving patient groups in the conversations around the design, actions and operations within the ecosystem and the cocreation of the patient platform and methods of data visualizations, any design features, tutorials, and even appropriate warning or explanatory messages. ‘Meaningful’ patient participation creates actionable patientcentred and patient-informed insights. Co-creation with PAB and their outreach is expected to result, via local/national-level governance and ethical systems, in patients’ views as guiding principles of the consortium. 7.1.3 Open and transparent science STRONG-AYA abides by the principle that the research pursued and insights generated using the data (including publications, statistics, software, etc.) and its disseminations shall be accessible to all stakeholders and the general public. Therefore our methods should be transparent and accessible, and knowledge developed should be shared and actionable across the ecosystem. Stakeholders commit to all algorithms used for data analytics being openly accessible and constantly reviewable/critiqued. All publications related to STRONG-AYA will be open access under relevant local or EU funding. Any dissemination activities are openly available to members of the wider public as well as specialists, via a register maintained by the STRONG-AYA Project Coordinator. While most insights and products will be accessible to members within the STRONG-AYA ecosystem, as well as the methods to produce these, the granularity to which the datasets can be accessed and what is queried STRONG-AYA – No. 101057482 – D5.5 will depend on the type of stakeholder requesting access and their motives. Therefore, while all methods and disseminations are open access, the individual level patient data will not be open access, adhering to data protection, and privacyand confidentiality-by-design principles. For equity in data provision and data queries, and hence to reduce potential for misalignment resulting from power asymmetry, the STRONG-AYA data ecosystem will ensure: • Data ownership and control will be held by local Principal Investigators and their teams. • The actions herein are not a new topology of data control exercise by a single partner. • Data analysis must still serve its named function if one partner conscionably opts out of a particular analysis or a specific use case (albeit in such cases on a likely reduced data sample size). Governance in STRONG-AYA STRONG-AYA partners believe that greater transparency about data ownership, collaborative analysis, and therefore clear data controllership promotes trust – among consortium partners and among stakeholders we disseminate results to. At local and national levels, STRONG-AYA governance is inclusive of local policy and legal agreements across the nations involved. Responsibility to adhere to these remains with the partners. Practically therefore, STRONG-AYA conforms to explicit levels of data anonymization; sufficient risk-guided population minimums must be met to permit analysis or re-analysis, to protect against unintended reidentification of individual data subjects (e.g. cohorts >10 as default, but may be adjusted with justification). At an international and pan-European levels, STRONG-AYA uses a self-governance model which is open and inclusive across its operations. Specifically, our principles encourage a wide community of experts to join in the production of scientific insights, accepting transparency in data, inclusiveness, inter-operability, collaboration, longevity, and sustainability. STRONG-AYA’s governance and legal principles align at the panEuropean, local and national levels. It aligns with, among others, the European Ethical Principles for Digital Health (2022), national and international privacy laws - such as the General Data Protection Regulation (GDPR; https://gdpr-info.eu/), and applicable sub-legislations relating to lawfulness; fairness and transparency; purpose limitation; data minimization; accuracy; storage limitation, integrity and confidentiality; and accountability. Within this Governance Model STRONG-AYA acknowledges and welcomes the differences between institutions and countries in their capacity, capabilities, and limitations in what data can or cannot be used in our analyses. There is an expectation that not all countries and institutions will be able to fully meet all requirements all the time (e.g. technical, administrative, legislative, etc.) and there will be gaps in policies as well as pragmatics of data security, quality control, and management. The ethos of a collaborative panEuropean ecosystem is that these gaps will impose less limits upon scientific learning within our data ecosystem. Therefore, a crucial task for STRONG-AYA to implement is the transparent communication and discussion of these potential differences and gaps, to allow room for their troubleshooting and secondary plans to achieve explicit goals. In the case of prospective (novel) data collection (such as PROs based on the COS), each local partner needs to ensure that transparency is present in communication with participants. This means that all participants have provided and documented informed consent. In the case of retrospective (historical) data available and provided by partners, article 4 of GDPR applies, thereby the data collected by a partner must meet all three grounds allowing the processing of such data. Namely; the data can be processed if a) the participant offered STRONG-AYA – No. 101057482 – D5.5 freely given, specific, and informed consent (Article 6 (1) (a)); b) processing is a legitimate interest exception such as there is a legitimate interest of controllers AND the legitimate interest cannot reasonably be achieved by using alternative means AND the interests and fundamental rights and freedom of the participant do not take precedence over the legitimate interests of the controller(s); c) the data is anonymized and therefore falls outside GDPR. As an example, a legitimate interest exception would be that of testing hypotheses for the purpose of public health. The Consortium encompasses a Committee for Coordination (CfC), which was established by NKI (as the leading organisation) to oversee the pan-European ecosystem. The CfC has a critical role in ensuring our principles are adhered to. Comprising leaders of local ecosystems and patient advocates it defines the responsibilities of members, including the Patient Advisory Board, defines and implements criteria for newly joining members, encourages adherence to founding principles and communication among national ecosystems. It also ensures consistency, standardization and harmonization of data and promotes scalability to and within countries to guarantee the sustainability and longevity of the network of ecosystems. The CfC as part of a defined organisational structure, has the responsibility to define and manage processes related to: • existing member actions, responsibilities, and rights; • new member access applications and rules of using of the data ecosystem; • aligning the acceptable motives and incentives for accessing and using the ecosystem; • defining and enforcing rules related to the submission of new use cases, utilising the existing ones and queries that can be run on the data that already exists in the ecosystem; • ensuring all members actively contribute and abide to the required standards; • resolving potential disputes and ambiguous situations; • sharing success, e.g. intellectual property from the use of the data ecosystem. As defined in our ecosystem goals (section 1.4), we have opted for a federated data ecosystem which preserves several ethical and governance objectives, without penalizing a relatively rapid pace of ‘learning’ (data analytics) that could be performed. It is important to clarify what is mean by ‘federation’ here. 7.2.1 Definition of federated learning ‘Federated learning’ refers to a group of computational methodologies where research questions or research problems (‘use cases’) that can be addressed mathematically and cooperatively, but without exchanging identifiable human subject-level data between any of the cooperating institutions (https://theodi.org/insights/explainers/what-is-federated-learning/). Our meaning of federation is NEITHER a cloud-based NOR a physically-located master data location, to which all partners must centralize their subject-level data to the ‘owner’ of that master repository on behalf of the consortium. In contrast, our concept of federation embraces fully decentralized digital data repositories – two or more of such repositories, ideally located inside each care-giving institution’s zone of control (where the individuallevel healthcare data was collected in the first place e.g. a hospital). In such as topology, the descriptive statistics and epidemiological mathematical modelling must necessarily be performed on multiple local computing devices, co-located within each data repository, on the sample size (e.g. > 10 cases per variable) that is contained inside each institution. STRONG-AYA – No. 101057482 – D5.5 Finally, the partial mathematical results (but most importantly NOT the subject-level data) from each institution are brought together on an encrypted telecommunications network, to generate the multiinstitutional overall ‘result’. For epidemiological models, this generally requires iterative and stepwise numerical integration of results, which converge on the mathematical ‘answer’. Technical details on the implementation will be provided in Section 3.5 later. The implementation of STRONG-AYA as a federated data ecosystem enables collaborative learning, and therefore aligns naturally with the following pan-European consortium aspirations: • Data Autonomy – our federation approach allows local and national ecosystems to remain fully in control of their data at all times, and no one outside of the institution can see their respective individual-level data, while at the same time cooperative overall insights and collective answering of research questions remains fully supported. • Data Solidarity – our AYA subjects altruistically gave consent to use their personal data to improve healthcare, but they do not want any negative repercussions to trace back to them leading to embarrassment, stigmatization, surveillance by authorities, or violation of privacy. By not exchanging any patient-level data, only statistical computations on multiple cohorts, this makes it extremely difficult for a malicious actor – either inside the consortium or outside – to re-identify any data subject even if there were successfully able to break the industry-grade telecommunications encryption. • Open Science – since decentralized computations are not physically performed by human eyes and hands, it is necessary to protocolize and pre-specify the statistical analyses a priori, which includes the selection of cohorts and the endpoints before analyses. Every data analysis needs to be explicitly written co-operatively among partners, as computer code which will be open access, permanently auditable and necessarily reproducible on public software repositories (such as GitHub) even though the patient data remains secret. • Diversity and Inclusivity – clinical and health care research benefit from learning as much as possible from the heterogeneity that is present in the real world. It is only by comparing and understanding differences in outcomes that the STRONG-AYA ecosystem and community improve the understanding the determinants of optimal value-based AYA cancer care. The ability to learn from sensitive personal data, without betraying the data subjects who have volunteered their data, helps us to address impacts of sample size, diversity and inclusivity when dealing with liminal health research subjects such as AYAs with cancer. In STRONG-AYA, no external persons are given any access to information about AYAs within local institutional repositories. Since patient data stays locally where they are initially generated, this facilitates opt-out for some types of analyses and enhanced opt-in for certain other analyses, provided that patient engagement and the COS-guided data collection can support this level of detailed ongoing engagement, without requiring modification of a centralised database. 7.2.2 Legal governance supporting federated data ecosystems A federated data ecosystem therefore requires, at minimum, a number of explicit agreements that establish: 1) the community of institutions and acknowledgement of their supplementary governance procedures – this is commonly known as a ‘Collaborative Research’ or equivalently ‘Consortium’ agreement. 2) Data protection impact assessments (DPIAs) that are resonant with interweaving of the local, national and international structural requirements. 3) A data access framework agreement that defines the principles of data controllership and permissible purposes for cooperative data processing in context of a federated STRONG-AYA – No. 101057482 – D5.5 telecommunications system. 4) Infrastructure service agreement(s) with disinterested technology service provider(s) that are able to host the technical and computational components, but do not have any say whatsoever about research questions, or patient data or knowledge that arise due to consortium activities. An example of agreement templates from previous federated learning projects, that have been used to develop the STRONG-AYA legal agreements, may be found openly accessible here: (https://www.medicaldataworks.nl/governance). In due course, the (redacted) STRONG AYA set of documentations will also be openly available here (www.strongaya.eu) as a template for future projects. Personnel in STRONG-AYA Local/national ecosystems consist of people with lived experience of AYA-onset cancer, personnel from multiple healthcare disciplines, including clinicians, allied health professionals, epidemiologists and individuals with data science skills. Each centre has its own responsible project and data managers. These, with the Principal Investigators, form the operational groups co-ordinating our local and partly national ecosystems. PPIE groups, charities and policy makers are integral to success, so each centre has clear plans and roadmaps for service user stakeholder engagement, as well as communication about the project with wider local healthcare professionals. We expect this to expand as other stakeholders, whose ethical principles align to those of STRONG-AYA, join our initiative. Data analytics technology and technical standards in STRONG-AYA In this section, we explore the technical nuances in more detail, specifically the key enabling technologies that makes a federated data ecosystem usable. Federated learning (‘FL’) is an established and rapidly maturing methodology having uniquely high value in pan-European projects, especially when dealing with sensitive personal health data. The legally-binding implementation of the European Health Data Spaces (EHDS) follows a long tradition of federation projects in Europe, that rely on novel privacy-enhancing technologies – for example, secure multi-party computation, differential privacy and federated learning – as a means of enhancing multi-institutional and global cooperation, while protecting the rights and confidentiality of data subjects. European-level projects of similar scale and construction as STRONG AYA include: (i) FeatureCloud, which uses encryption and blockchain-based computations to protect individual identity. (ii) TRUMPET, which collectively learns from multi-site radiation dosimetry data in order to personalize radiotherapy treatment. (iii) AI4EOSC, which will provide computationally intensive federated Artificial Intelligence to researchers in the European Open Science Cloud. (iv) DIGIONE, which pioneers an EU-wide rapidly learning oncology system aiming to optimize treatment selection for cancer patients. We have highlighted above the roles of local institutions, national bodies and international consortia in the formation of a federated data ecosystem. To connect these with an easily visualizable technology, we may consider: (i) the local institutional and national data repositories as stand-alone ‘stations’. (ii) instead of moving personal patient data around, we will instead send out computer code that can perform statistical computations on our behalf, i.e. ‘trains’. (ii) while the data is inherently protected by its own data stations, the integrity of the telecommunications connectivity between stations must be kept private to the consortium, and safeguarded against malicious actors trying to ‘eavesdrop’ on the statistical results or in STRONG-AYA – No. 101057482 – D5.5 some way attempt to re-identify one of the data subjects. Such protections had to be woven into the digital and operational infrastructure of the ecosystem, which we would call ‘tracks’. This gives us the easily accessible analogy of protocolized ‘trains’ travelling outwards to privately held data ‘stations’ to do statistical analysis, then bringing together the cohort-based analysis results (no individual data) using cybersecurity-assured ‘tracks’, i.e. an implementation of the “Personal Health Train” paradigm (https://www.dtls.nl/fair-data/personal-health-train/). 7.4.1 Autonomy of local and national data collection structures Where local systems already collect Patient-reported outcome or Patient-reported experience data, this is utilised for STRONG-AYA data collection, and where there are not such local systems the expertise in the consortium supports its implementation. That is, each participating institution retains full control of its data repository, its data acquisition resources and its own data management policies, using insofar as possible the technical capabilities it has built up prior to forming STRONG-AYA. However, to support systematic and internationally harmonized collection of AYA data, the Consortium has now published and deployed the ‘COS’ for use at the local and national levels. 7.4.2 Technical implementation of data ‘stations’ Whilst local and national actors fully control and entirely own their respective data, some intermediation technologies are required by each consortium partner to make this data simultaneously private yet also available for collaborative learning. Therefore STRONG-AYA develops, extends and deploys for all partners a set of free, open source and publicly accessible software tools for bridging institutional data formats with the consortium COS, despite the range of local and national systems 32. In STRONG-AYA, we adhere to the Findable-Accessible-Interoperable-Reusable (‘FAIR’) data management principles 33. This states two key principles that must be emphasized : (i) FAIR data does NOT imply that data is openly accessible, and (ii) that FAIR-ness constitutes data which is suitable – by way of ‘metadata’ annotations describing the actual data - that makes it possible for both humans (manually) and computer code (remotely) to interact with the data. Additionally, the FAIR principles states that not one single format or structure necessarily takes precedence over any other. For example, the OMOP Common Data Model (‘CDM’) is popular in many countries, but STRONG-AYA does not exclude any member’s data solely on the basis that it did not conform to one ‘CDM’. We include a plurality of data systems, structures and formats, already in use in local and national systems, some more and some less interoperable with each other. The key advantage of a ‘metadata overlay’ type of methodology is that, as a consortium, we will strive for a standardization of an optimum minimal level of data, and we can harmonize/inter-calibrate the rest, thus aiming at syntactic and semantic interoperability that will mature in parallel with the Consortium. Whenever the Consortium needs to enhance the COS or update the definitions, it is only necessary to revise the metadata, not the actual patient data itself. Finally, multiple code dictionaries co-exist within the metadata, so the metadata implements the harmonization across terminologies and measurement scales, and it is not necessary to revise or re-code the raw data, to ‘force’ equivalencies, after it is collected. Federation also supports decentralized statistical model ‘training’ (e.g., regression, neural networks) without compromising data privacy and adhering to data minimization thresholds. STRONG-AYA – No. 101057482 – D5.5 7.4.3 Statistical computing code as standalone ‘trains’ All code for use by the Consortium is placed in a publicly readable repository, such that any person can openly examine the code and inspect it for potential vulnerabilities, and local data experts are required pragmatically to review that code as it is created. For consortium work, technology partners are included, and independently cross-verify that the code performs the mathematical function correctly, and does not contain any malicious action, before the specific version of the code in the repository is locked for use within the consortium. Only locked code is packaged into standalone, independent, and operating-system agnostic software applications (i.e., Docker® containers) which we refer to as ‘trains’. The train thus encode the statistical analysis requested and desired by the Consortium partners acting jointly, for example, one train may contain a generalized regression model, or another train may have code to statistically summarized prevalence of different cancer types. Thus development of the trains goes hand in hand with: (i) the specific use cases expressed by the Consortium, (ii) the agreement in advance of the Consortium about which data fields and what analysis needs to be performed, and (iii) how the final results of the computational actions are to be made available, and to whom. 7.4.4 Data analytics dashboards The statistical results and epidemiological models coming from the STRONG-AYA data collections needs to be presented in a cognitively accessible way, to be disseminated to champions who can then implement transformation of AYA care in their local and national ecosystems. We take this to mean ‘dashboards’ (i.e. user-friendly data visualization web-browser based applications). These present the results of, or insights from, analyses completed upon the consortium data either on a scheduled (time-based), ad hoc (e.g. a specific one-off clinical hypothesis test) or on-demand (e.g. a user wishes to see a generalized cohort summary of PROs values across broad range of cancer types immediately, for good reason). Therefore, co-design and co-creation – meaning clinicians, researchers, AYAs and policy makers as a cohesive whole – is fundamental to the way we report our results. The underlying technology is intended to be invisible and unobtrusive where possible, while the user focusses on gaining insight, communicating knowledge and applying figures/graphs created. The aim is to integrate AYA-specific aspects into, for example, research questions (from EuroCare enquiries), patient-driven questions (from AYA discussion fora) and for policy drafting (such as by ECO). The objective here is to make data insights, such as the COS accrued by our Consortium, accessible in a specific time-frame including ‘in real time’, and with granularity befitting the user and the question. The choice of web-browser based application as a design principle is informed by the need to partly decouple the communication/dissemination aspects, which require capabilities in web design and visual user interaction, from the underlying infrastructure technology which handles the privacy, access controls, network cybersecurity and telecommunications protocols. The former is jointly a work in progress with consortium partners including YCE; the latter has been contracted in large part to a Medical Data Works B.V. that has specific expertise in Vantage6 open-source federated infrastructure. STRONG-AYA – No. 101057482 – D5.5 7.4.5 Privacy enhancements in STRONG-AYA A federated approach for AYA-onset cancer needs to address the challenges of co-ordinated working with small case numbers and yet provide timely, richer and more robust insights into care or treatment compared to single database or formal data transfer methods. In STRONG-AYA our greatest shared responsibility is data security. This begins first at the local ecosystem level, where the detailed datasets are collected and stored. At all levels across STRONG-AYA, access and control protocols have been determined and developed to block certain information reaching specific stakeholders. For example (in future) a pharmaceutical or medical devices commercial partner may not be permitted to develop analyses involving specific drugs administered or special procedures carried out; however such information might be made accessible to clinicians and researchers who wish to develop prediction models. Well-developed access management from local levels shall underpin cooperation at an international level to support the beneficial use of the consortium infrastructure. The privacy enhancement in federated learning is underpinned by the unfeasibility of reverse-engineering a particular participants’ identity from summaries, statistical models or other such combined analysis results based on groups of participants (typically presented in reports and publications). Second, our data security must consider statistical analyses requiring frequencies of outcomes to answer research questions – such as the numbers of persons ‘alive (or deceased)’, ‘diagnosis positive (or negative) or a time interval (such as ’90 days since a certain event’). In special types of research questions, there is a risk that such an outcome may be characteristic to only one or a few patients – in which case the descriptive statistics or epidemiological model may no longer obscure the data values of individual subjects. OUr infrastructure is designed to block analyses that do not have sufficient cohort size. By default, as stated earlier, this number is arbitrarily set at 10, however institutions in certain use cases may have justification to increase or reduce this number, on a case-by-case risk-guided basis. Finally, the development of the COS itself is based on the idea of data minimization, even though no personlevel data is being shared. Structured information from the institutional collection of the COS data fields will be extracted by investigators within each institution, then populated into their respective STRONG-AYA data ‘stations’ controlled by their own institution and governed by their own data protection regulations. All data placed in these areas will be checked to ensure it is pseudonymized or anonymized (as consistent with the DPIA and data access agreements) to avoid the re-identification of individuals, and to remain compliant with GDPR and local regulations. Further, potentially traceable identifiable information which are not necessary for consortium analyses - such as birth dates and postcodes – are removed after extraction, and all absolute dates of clinical events will be replaced by time intervals, say number of days from diagnosis till recurrence of cancer, for example. 7.4.6 Challenges encountered in the federated analytics approach While federated learning comes with certain attendant advantages in terms of data access and transparent analytics, there are additional considerations that must be acknowledged. Federated ecosystems may be able to process disparate, large, and complex datasets, to accelerate findings in a specific health field and thereby increase research power, detail, and reach. The federated data ecosystem concept acknowledges and respects that data collection workflows and data management practices will be specific locally, and already in use extensively among local project partners for numerous operational and research purposes. Datasets owned by project partners will be re-used for the unitary common goal of answering pre-defined clinical research questions, through initial priority research questions. STRONG-AYA – No. 101057482 – D5.5 However, this autonomy and control over data at the local and national levels will also introduce personnel, expertise and resource allocation questions for the institutions that participate in the federated data ecosystem. In centralized designs, the partners may – if they wish – take a less active role by sending data to the principal investigator, thereby relying on computing resources, data cleaning, domain expertise and analysis skills available at the leading institution. The federated approach warrants that the skills and expertise are more dispersed among the partners, rather than being concentrated in one. There is therefore a strong and significant case for consortium-wide technical support, knowledge sharing, and deployment of customizable software tools to facilitate working in a federated fashion. Additionally, there needs to be a fundamental willingness from the lead institution to share its expertise and support the maturation of research capacity (along with its attendant resources) more widely around the consortium, rather than keeping these in one location. 8 What can we learn from similar initiatives? The Health Outcomes Observatory (H2O) project is a public-private partnership funded by the European Union. It aims to build an international ecosystem to incorporate health outcomes, including patientreported outcomes in healthcare decision making across three diseases: diabetes, cancer and inflammatory bowel disease. There are several similarities with STRONG AYA. Both projects are interested in outcome sets, particularly patient-reported ones, and both use a federated data infrastructure and seek to use digital tools to involve patients in the data sharing pathway. They differ in terminology; H20 has favoured the term ‘observatories’ to describe its data infrastructure whereas we use the term ‘ecosystem’ and they vary in their governance structure. H20 has opted to create separate non-for-profit legal entities, co-led by a public and private enterprise representative in each country. These entities are currently operational in three of the four countries involved in the project: Austria, Spain and the Netherlands (Germany has not yet been set up according to public resources). A review of their website (https://health-outcomes-observatory.eu/) suggests that this legal entity route may allow for further grant acquisition later as they would be new entities applying for funding, but that a considerable amount of time and resources were allocated within the project to the bureaucracy to create these entities. Taking these steps at an early stage potentially limits buy-in from other parties as feasibility has not yet been proven and buy-in is based on the premise rather than the actual function. Furthermore, the national observatories may be national in their legal registration but face the challenge to produce nationwide data if they cannot on board multiple centres, data sources, or nation-wide studies in each country. The Dutch national observatory still only includes the Erasmus University Medical Centre as a source of data, and similarly only one medical centre is involved in Spain and in Austria and Germany, where one centre is contributing data but no national observatory has been established. No panEuropean observatory has been formally established, although international aggregated data is being used and visualized via the H20 Insight Centre (https://public.tableau.com/app/profile/k.funk.tableau/viz/H2OInsightsCentre/H2OInsigtsCentre), a public facing dashboard. Patient recruitment also currently sits at under 1100 patients according to the Insight Centre statistics. By defining our ecosystems by data source and by technical functionality and with flexible pragmatism, STRONG AYA’s ecosystems can perform earlier analysis on a larger number of patients to generate insights earlier. Data for the Common Good (https://commons.cri.uchicago.edu/) is an example of a strong and effective data insights ecosystem, with related but distinct objectives and methods. Its focus is across diseases, but encompasses cancer. Its reach is intercontinental, with a particular emphasis on diseases affecting young children, though it features an AYA cancer working group. The initiative prioritises existing data (in particular, STRONG-AYA – No. 101057482 – D5.5 from a community of experts to help refine it further. Training programmes for healthcare professionals on the technical infrastructure and on the use of PROMs in research and clinical practice might also help generate greater buy-in and support for systems that use patient reported outcomes. Longevity/sustainability The STRONG-AYA approach is adaptable to several changes in technology – our use of Application Programming Interfaces ‘APIs’ for example is accepting of the inevitable new software implemented in healthcare centres. The technology used in STRONG-AYA is itself scalable but is dependent on the resources of partners wishing to get involved and their financial and political constraints. Here the flexible approach of STRONG-AYA to overcome these differences will be paramount to longevity, maintaining inclusivity and quality of analysis. The development the COS for AYAs with cancer enables STRONG-AYA to offer new opportunities for the collection of homogeneous prospective data, which will support new research and healthcare innovation. The agile and adaptable approach of FL allows flexibility in incorporating existing systems and practices in larger ecosystems, allowing for the accommodation of new local contexts and ecosystems. Tools which will enable and motivate stakeholders to join and work together are the websites, specific ways of set-up of local databases, portals for stakeholders to enter and access desired information, and finally the access to the expertise required to operate ecosystems at local and European levels. Finally, STRONG-AYA will remain responsive to emerging and unanticipated ethical issues by adhering to iteratively developed ethical guidance throughout the duration of the project, including patient-centred. Close links, communication and collaboration with user groups and stakeholders are vital to ensuring that STRONG-AYA evolves with their needs and as new research questions arise in AYA. As a Consortium the inclusion of early career researchers across all disciplines will help ensure that ongoing success and longevity of STRONG-AYA. 10 Future actions and considerations There are a range of future actions under consideration: Strong-AYA can grow and expand for example by - Engaging other local or national data ecosystems about AYA-onset cancer. - Incorporating other types of data e.g. genomics, pharmacological data - Widen functions and use cases, and adapting where needed within our governance framework At the heart of this are the young people with lived experience of cancer. The STRONG-AYA ecosystem primacy of continued patient involvement recognises that technical solutions are only one part of the advances. The next steps need to focus on scaling and sustaining meaningful PPIE and co-design. As those with lived experience learn with the project, they become better advocates and invaluable peers in the Consortium, contributing their expertise across various domains, from research and data analysis to marketing techniques. STRONG-AYA is uniquely positioned to empower AYAs with personalized, tailored information and opportunities to explore their own data, all the while enhancing digital health literacy—a skill crucial for the 21st-century. Through time, this can result in ever more seamless integration or research STRONG-AYA – No. 101057482 – D5.5 into routine clinical care, so that participation benefits individuals, society, as well as medical communities. Achieving these goals requires respecting and expanding diversity of patient perspectives, working transparently, and prioritizing ethics, privacy, and security. 11 References 1. Group AaYAOPR: Closing the gap: Research and Care Imperaives for Adolescents and Young Adults with Cancer, in NIH (ed): No. 06-6067. Bethesda, MD, National Institute of Health, National Cancer Institute, and Livestrong Young Ault Alliance, 2006 2. Close AG, Dreyzin A, Miller KD, et al: Adolescent and young adult oncology-past, present, and future. CA Cancer J Clin 69:485-496, 2019 3. Gupta S, Harper A, Ruan Y, et al: International Trends in the Incidence of Cancer Among Adolescents and Young Adults. J Natl Cancer Inst 112:1105-1117, 2020 4. Bray F, Ferlay J, Soerjomataram I, et al: Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 68:394-424, 2018 5. Lewis DR, Seibel NL, Smith AW, et al: Adolescent and young adult cancer survival. J Natl Cancer Inst Monogr 2014:228-35, 2014 6. Trama A, Botta L, Foschi R, et al: Survival of European adolescents and young adults diagnosed with cancer in 2000-07: population-based data from EUROCARE-5. Lancet Oncol 17:896-906, 2016 7. Adams SC, Herman J, Lega IC, et al: Young Adult Cancer Survivorship: Recommendations for Patient Follow-up, Exercise Therapy, and Research. JNCI Cancer Spectr 5, 2021 8. Chao C, Bhatia S, Xu L, et al: Chronic Comorbidities Among Survivors of Adolescent and Young Adult Cancer. J Clin Oncol 38:3161-3174, 2020 9. Palmer JD, Hall MD, Mahajan A, et al: Radiotherapy and Late Effects. Pediatr Clin North Am 67:1051-1067, 2020 10. Ryder-Burbidge C, Diaz RL, Barr RD, et al: The Burden of Late Effects and Related Risk Factors in Adolescent and Young Adult Cancer Survivors: A Scoping Review. Cancers (Basel) 13, 2021 11. Woodward E, Jessop M, Glaser A, et al: Late effects in survivors of teenage and young adult cancer: does age matter? Ann Oncol 22:2561-2568, 2011 12. Fidler MM, Frobisher C, Hawkins MM, et al: Challenges and opportunities in the care of survivors of adolescent and young adult cancers. Pediatr Blood Cancer 66:e27668, 2019 13. Perez GK, Salsman JM, Fladeboe K, et al: Taboo Topics in Adolescent and Young Adult Oncology: Strategies for Managing Challenging but Important Conversations Central to Adolescent and Young Adult Cancer Survivorship. Am Soc Clin Oncol Educ Book 40:1-15, 2020 14. Ferrari A, Stark D, Peccatori FA, et al: Adolescents and young adults (AYA) with cancer: a position paper from the AYA Working Group of the European Society for Medical Oncology (ESMO) and the European Society for Paediatric Oncology (SIOPE). ESMO Open 6:100096, 2021 15. Collaborators GBDAYAC: The global burden of adolescent and young adult cancer in 2019: a systematic analysis for the Global Burden of Disease Study 2019. Lancet Oncol 23:27-52, 2022 16. Bleyer A, Barr R, Hayes-Lattin B, et al: The distinctive biology of cancer in adolescents and young adults. Nat Rev Cancer 8:288-98, 2008 17. Dommett RM, Redaniel MT, Stevens MC, et al: Features of cancer in teenagers and young adults in primary care: a population-based nested case-control study. Br J Cancer 108:2329-33, 2013 18. Fern LA, Birch R, Whelan J, et al: Why can't we improve the timeliness of cancer diagnosis in children, teenagers, and young adults? BMJ 347:f6493, 2013 STRONG-AYA – No. 101057482 – D5.5 19. Meeneghan MR, Wood WA: Challenges for cancer care delivery to adolescents and young adults: present and future. Acta Haematol 132:414-22, 2014 20. Levin NJ, Zebrack B, Cole SW: Psychosocial issues for adolescent and young adult cancer patients in a global context: A forward-looking approach. Pediatr Blood Cancer 66:e27789, 2019 21. Keegan THM, Abrahao R, Alvarez EM: Survival Trends Among Adolescents and Young Adults Diagnosed With Cancer in the United States: Comparisons With Children and Older Adults. J Clin Oncol 42:630-641, 2024 22. Smith AW, Seibel NL, Lewis DR, et al: Next steps for adolescent and young adult oncology workshop: An update on progress and recommendations for the future. Cancer 122:988-99, 2016 23. Osborn M, Johnson R, Thompson K, et al: Models of care for adolescent and young adult cancer programs. Pediatr Blood Cancer 66:e27991, 2019 24. Fardell JE, Patterson P, Wakefield CE, et al: A Narrative Review of Models of Care for Adolescents and Young Adults with Cancer: Barriers and Recommendations. J Adolesc Young Adult Oncol 7:148-152, 2018 25. Das M: New report from STRONG-AYA on addressing needs of adolescents and young adults with cancer in Europe. Lancet Oncol 25:960, 2024 26. Shah A, Stewart AK, Kolacevski A, et al: Building a Rapid Learning Health Care System for Oncology: Why CancerLinQ Collects Identifiable Health Information to Achieve Its Vision. J Clin Oncol 34:756-63, 2016 27. Lambin P, Roelofs E, Reymen B, et al: 'Rapid Learning health care in oncology' - an approach towards decision support systems enabling customised radiotherapy'. Radiother Oncol 109:159-64, 2013 28. Price G, Mackay R, Aznar M, et al: Learning healthcare systems and rapid learning in radiation oncology: Where are we and where are we going? Radiother Oncol 164:183-195, 2021 29. Cromie KJ, Crump P, Hughes NF, et al: Data Resource Profile: Yorkshire Specialist Register of Cancer in Children and Young People (Yorkshire Register). Int J Epidemiol 52:e18-e26, 2023 30. Husson O, Ligtenberg MJL, van de Poll-Franse LV, et al: Comprehensive Assessment of Incidence, Risk Factors, and Mechanisms of Impaired Medical and Psychosocial Health Outcomes among Adolescents and Young Adults with Cancer: Protocol of the Prospective Observational COMPRAYA Cohort Study. Cancers (Basel) 13, 2021 31. Vlooswijk C, Poll-Franse LVV, Janssen SHM, et al: Recruiting Adolescent and Young Adult Cancer Survivors for Patient-Reported Outcome Research: Experiences and Sample Characteristics of the SURVAYA Study. Curr Oncol 29:5407-5425, 2022 32. Gouthamchand V. CA, Hoebers FJP., Wesseling FWR., Welch M., Kim S., Kazmierska J.,, Dekker A., Haibe-Kains B., van Soest J., Wee L.: Making head and neck cancer clinical data FindableAccessible-Interoperable-Reusable to support multi-institutional collaboration and federated learning. BJR|Artificial Intelligence 1:ubae005, 2024 33. Wilkinson MD, Dumontier M, Aalbersberg IJ, et al: The FAIR Guiding Principles for scientific data management and stewardship. Sci Data 3:160018, 2016 34. Prainsack B, El-Sayed S: Beyond Individual Rights: How Data Solidarity Gives People Meaningful Control over Data. Am J Bioeth 23:36-39, 2023 35. Prainsack B, El-Sayed S, Forgo N, et al: Data solidarity: a blueprint for governing health futures. Lancet Digit Health 4:e773-e774, 2022 36. S. Oliveira, M.I., Barros Lima, G.d.F. & Farias Lóscio, B. Investigations into Data Ecosystems: a systematic mapping study. Knowl Inf Syst 61, 589–630 (2019). https://doi.org/10.1007/s10115-018-13236