Member State & EU level priorities

Based on the feedback on the six domain areas identified as levers for change, Round Table Meeting participants considered the key Member State and EU level priorities.

For further information on the domain discussions that underpinned the decisions on these priorities, visit the relevant rooms in the Hub or download the summary report.

Priorities to be driven at a Member State level

1. Improving knowledge, education and training

A greater focus on the comprehension of data science and artificial intelligence (AI) is needed in certain parts of the healthcare system. In addition, there needs to be clear, accessible information about the use of data and AI, specifically tailored for patients and citizens. As part of this effort, clinical leaders need to become knowledgeable champions of AI, help communicate its benefits to support a wider understanding, and provide positive role models in what is often a hierarchical healthcare culture.

Healthcare professionals (HCPs) at all levels need further training regarding basic digital literacy on one hand, but also skills training in order to learn how to actually use and integrate new technologies into their working routines. Finally, a change of attitude towards innovation has to be fostered. This requires HCPs who can work alongside many different professions and roles, including those in AI-driven technologies. Action to drive this initiative lies at the Member State level (universities, etc.) however EU funding will be needed to achieve it.

The upcoming Digital Europe programme will provide a total of €700 million to support the development of advanced skills, including knowledge of AI. As part of the Coordinated Plan on Artificial Intelligence, Member States and the European Commission have agreed to work together to develop materials that can be used in awareness campaigns regarding the benefits of AI.

2. Establishing national networks

Several Round Table Meetings suggested it would be beneficial to establish national bodies or networks that would work across domains. These might be similar to, or even build upon, the existing Digital Innovation Hubs. The objectives would be to boost the exchange of data and expertise, encourage seamless integration of clear standards, certification, innovation efforts across all Member States and provide support for regulatory bodies regarding the implementation of new rules and policies.

One such network is CLAIRE – the Confederation of Laboratories for Artificial Intelligence Research in Europe – which aims to strengthen European excellence in AI research and innovation across Europe, with a human-centred focus. The CLAIRE initiative intends to establish a pan-European network of Centres of Excellence in AI, strategically located throughout Europe, with a central state-of-the art facility. The planned lighthouse initiatives of the European Commission, which bring together centres of research, innovation and expertise, may also have a role here.

3. Developing new financial models

New value-based healthcare financial models are needed within national healthcare systems to support the validation and deployment of AI driven innovations. These should be outcomes based, taking into consideration the entire patient pathway. This may require broader restructuring of payment models, integrating both primary and secondary care. There will also need to be recognition that some AI applications are not replacements for existing processes or systems but additions. While these new innovations may not necessarily generate immediate costs savings, they may provide longer-term system benefits.

Priorities to be driven at an EU level

1. Building a robust data infrastructure for Europe

Participants at national Round Tables Meetings expressed the urgent need to establish a secure, trustworthy and competitive cloud infrastructure in Europe that would benefit public administration, businesses and citizens.

It was recognised that there are ongoing discussions and projects within Europe that have similar objectives, which might provide opportunities for driving AI in healthcare:

  • The European Commission, in collaboration with Member States, is currently undertaking preparatory work towards development of the European Health Data Space, an infrastructure that will facilitate the sharing of health data for public health, treatment, research and innovation across the EU.
  • GAIA-X, a project being driven by representatives from the politics, business and science sectors in France and Germany, together with other European partners. This initiative aims to develop a secure, federated data infrastructure for Europe that meets the highest standards of digital sovereignty while still promoting innovation. Its objective is to develop an open, transparent digital ecosystem, where data and services can be made available, collated and shared in a trusted environment.
  • In October 2020, the European Commission announced that all 27 EU Member States had committed to working towards developing a secure, next-generation cloud for Europe to provide interoperable, pan-EU, trustworthy data processing infrastructure and services for the public and private sectors. Such an infrastructure might be something that could support the hosting and sharing of health data across Europe.
  • In September 2020, as part of the ‘Digital Decade’ agenda, the European Commission proposed a new regulation for the European High-Performance Computing Joint Undertaking. This aims to advance Europe’s leading role in supercomputing and quantum computing which will underpin the overall digital strategy, including big data analytics, AI, cloud technologies and cybersecurity. This initiative could boost scientific breakthroughs with AI in the healthcare sector.

2. Providing guidance on data management and governance

Feedback from the Round Table Meetings called for clear guidelines and common standards on data management strategies, including requests for data, data collection, infrastructure and maintenance, storage, access, anonymisation, governance, security and business models for financial sustainability of data repositories. While many such standards already exist across Europe, there is currently a lack of uptake of these tools; consequently, barriers to their adoption need to be explored.

Progress is already being made within the EU:

  • The European Commission has recently adopted a Recommendation on a European Electronic Health Record (EHR) exchange format which will facilitate cross-border interoperability. This sets out a framework for further developing a format that will enable citizens to securely access their health data and for it to be shared with HCPs across borders in the EU. The recommendation will include a set of common technical specifications for the exchange of data and set out the principles that should govern this exchange, such as ensuring data protection and security, in line with General Data Protection Regulation (GDPR) and full compliance with the cybersecurity framework.
  • EHAction, which supports the European Commission’s eHealth Network, aims to facilitate the sharing of health data across borders, to improve interoperability and cybersecurity, while ensuring privacy and data protection requirements.
  • The Ethics guidelines for trustworthy AI produced by the European Commission’s AI HLEG along with the final assessment list for trustworthy Artificial Intelligence in July 2020 will help support the safe and transparent development of AI applications. The European Commission’s white paper has furthermore suggested that these guidelines may be used for training purposes and to establish voluntary certification schemes.
  • In November 2020, as part of the 2020 European strategy for data, the European Commission proposed rules for a Data Governance Act. This aims to improve data availability for use by increasing trust in data intermediaries and by strengthening data-sharing mechanisms across the EU. The act calls for the creation of a European Data Innovation Board (EDIB), an expert group that will ensure consistent practices in the processing of requests for data and regarding the general authorisation framework for data sharing services. The EDIB will provide support and advice to the European Commission and be composed of expert representatives of Member States as well as representatives of the different sectors and common European data spaces.

3. Providing guidance on regulation and risk assessment of AI solutions

The Round Table Meetings highlighted that guidance is needed from the EU regarding best practice in the development and regulation of AI solutions to encourage streamlined application and combat fragmentation.

Secondly, clarity is required as to how the existing EU Medical Device Regulation (MDR) should be applied to the many different types of AI application. Questions surrounding the specificities of AI surfaced in all Round Table Meetings. Regulation must make adequate an assessment about the type of AI system in question, and whether AI should be treated as a separate component. More clearly, the definition of AI from a regulatory perspective must be clarified and the distinction between any issues caused by AI or other components outlined.

The ‘care versus cure’ distinction that emerged from the Round Table Meeting in the Netherlands is important. Indeed, any future regulatory framework, risk assessment or policies must bear in mind whether the AI system in question is used to directly affect a patient’s life or to alleviate non-critical workload of healthcare personnel. As such, proposed measures across the EU must be coherent as well as sufficiently flexible and nuanced.

It is clear that regulation should not hinder innovation but rather encourage it. To that end, it will likely benefit from being supplemented with standardisation and certification efforts. Common European standards are required that can be adopted at Member State level. There are many large organisations that could be involved in these developments and implementation, and where expertise could be harnessed, such as the International Organisation for Standardisation (ISO) and the World Wide Web Consortium (W3C). Beyond that, the European Commission has suggested that a self-certification or labelling scheme for trustworthy AI applications might be a promising incentivisation model across Europe to encourage the development of beneficial AI systems.

In all of these cases it may benefit Member States to create their own expert national bodies to drive regulation at a local level. These might equally be supplemented by expert advisory committees or groups, or more granularly by ‘innovation ethics committees’, supervising clinical trials in hospitals which involve AI systems.

Furthermore, a dedicated body representing multiple stakeholders and financed by the EU could support the establishment of the European Health Data Space to independently curate data, and navigate challenges and implementation across Member States. This body could equally work on defining and clarifying the legal framework for data usage when it comes to the use of patient data for the benefit of society. As suggested at the Round Table Meeting in Ireland, this could also be a government-led body within a Member State acting as a guardian of health data and connected to others in a network across the EU.

With regard to regulation, the European Commission is currently developing proposals for a new legal framework outlining the ethical principles and legal obligations to be followed when developing, deploying and using AI, robotics and related technologies in the EU, including software, algorithms and data. The legislative proposal is expected to be published by the European Commission during 2021.

Multiple Round Table Meetings pointed out that Member States are already evaluating how to adapt existing bodies for the purpose of monitoring and regulating AI, and, for example, the French data protection authority, CNIL, has already adopted standards for health sector data relating to the processing and retention of personal data, including that used for research, study, and analysis in the health sector.

Regulation, risk assessments, testing and experimentation as well as certification and standardisation efforts cannot exist within a vacuum. As previously highlighted, they must be firmly grounded within a novel infrastructure across the EU. To that end, the European Commission’s white paper has suggested that the ecosystem within Europe will be supported by AI-on-Demand platforms, a public–private partnership on AI, data and robotics in the context of Horizon Europe, and to facilitate the adoption of AI in healthcare, underpinned by a dedicated ‘Adopt AI Programme’, supporting public procurement of AI.

Moreover, testing facilities and regulatory sandboxes might be useful to establish across Europe for the initial testing and assessment of AI systems and to inform regulation and standardisation of the capabilities of AI systems on an ongoing basis. Regular monitoring of capabilities should be continued post deployment.