Thoughts from the co-chairs

Here the co-chairs share their thoughts on some of the key topics of discussion from the Round Table Series and how these may shape the way forward for the successful adoption of artificial intelligence (AI) in healthcare in Europe.

Farzana Rahman
CEO, Hexarad UK
Follow on LinkedIn

AI is already part of our day-to-day lives, whether it’s the spam filters on our email, friend suggestions on social media or voice assistants, such as Alexa. Yet when we look at the use of AI within healthcare, we can see that its adoption has been slower than in other sectors. This is perhaps inevitable; healthcare deals with personal, sensitive information at times when we are at our most vulnerable. This has led to debates about the impact of AI within healthcare, with concerns being raised about the impact on workforce, ethical issues and the use of personal data.

However, aside from this, there are also very real organisational factors which can impede technological innovations such as AI within the healthcare setting. Healthcare providers, start-ups, and research centres often work in silos, not communicating with each other. Products that are developed without early input from clinical users may have limited efficacy resulting in poor uptake. Collaborative working between these organisations is key to addressing this problem. This could also help address the challenge of resource distribution and scalability which can be a particular issue when looking at health systems in rural versus urban areas. Collaborative working will require alignment between both the people and technology within organisations, with interoperability and data-sharing being important factors to consider.

As healthcare organisations consider how they can use AI, they should also think about the investment in people and resources that will be needed to manage this change. This is especially important in healthcare where the stakes are high and there are serious consequences to getting things wrong. This requires strong leadership by decision makers, with an understanding that any investments in technology must be accompanied by investment in people.

It is also important for leaders to understand that improvements in efficiency and costs may take time to be realised and that the goal should be long-term improvement rather than short-term gain. By focusing on tools that support clinicians, such as those which reduce administrative burden, organisations can ensure buy-in and support for the use of these solutions from their staff.

Governments will need to think about funding mechanisms and how these could affect the uptake of AI within organisations. The reimbursement of AI remains complicated in Europe, with national and local payors often sharing responsibility for this. Consideration of reimbursement pathways, together with financial incentives for innovation, will be key when planning large-scale roll out of AI tools within healthcare.

Zineb Nouns
Clinician, medical education specialist and HR manager
Follow on Twitter

AI is fundamentally changing healthcare as we know it. AI offers the opportunity to boost healthcare services by: increasing the speed of new developments, improving medical quality, freeing up working hours and ultimately, lowering costs. However, this development also presents quite a challenge for people working in the field:

• a fear of being ‘replaced by AI’
• the need for learning and unlearning, which can be frustrating
• a lack of systems connectivity that puts a high burden on time-consuming and often redundant ‘machine feeding’ activities
• ethical and data privacy concerns remain unresolved

However, the progression of the integration of AI into healthcare is impossible to ignore. Therefore, we have to find ways to empower and enable our workforce to adopt AI and also to provide the expertise for its further development.

AI and automation will have a tremendous impact on the healthcare workforce. Not only will the established professions change in their specific roles, but entirely new professions will emerge. In particular, where medical roles and data science intersect, new professions – for example AI engineers – need to be defined, established, trained and integrated.

All healthcare professionals (HCPs), both established and newly emerged, will have to learn to work with one another as well as with new AI applications. The approach to patient communication will change around those new developments too. The educational challenge is far reaching – from a change of mindset, culture, skills and everyday behaviour to ultimately the establishment of entirely new workflows and medical practices.

Education and training curriculums need to address the necessary shift in priorities. For example, physicians should no longer graduate from university without a basic technological education. Similarly, it is no longer acceptable that nursing students devote more time to learning how to make a bed than how to use digital hospital systems or AI-supported processes.

The ability to work in interdisciplinary rather than hierarchical teams, as well as communicating with patients who nowadays have much easier access to substantial medical knowledge, will become even more crucial skills than they are today.

Charlotte Stix
Former coordinator for the European Commission’s high-level expert group on artificial intelligence
Follow on Twitter

AI holds a host of promises to help citizens live longer and healthier lives. At the same time, the introduction, development and deployment of this technology must be well considered, trade-offs weighted accordingly, and steps taken to ensure safe, robust and trustworthy AI applications. While this holds for all sectors, it is especially crucial in the healthcare sector where AI is set to change how we prevent and tackle disease. Because of the sensitive nature of AI within this sector, it is vital to encourage diverse and inclusive multi-stakeholder processes by harnessing existing knowledge and highlighting opportunities as well as concerns.

The discussions and recommendations resulting from the EIT Health Round Tables were insightful and it is crucial that they are implemented and actioned, both on Member State and pan-European level.

First and foremost, it must be noted that many pointed out that the regulation of AI is challenging, especially when it comes to the nuances of the healthcare sector. The scope, depth and flexibility of any future regulation should take into account the opinions and concerns of a broad range of stakeholders, from patient organisations to clinicians and more, in order to truly account for the needs and hesitations of those most affected. To that end, specific consultations and analyses could be undertaken, or multi-stakeholder alliances, expert advisory boards and cross-sectoral committees established, as has been suggested in the Round Tables. All this can serve to ensure that any regulation is adequately flexible whilst also being sufficiently precise.

To ensure informed regulation and policy, the knowledge gap that currently exists between various actors in the space needs to be addressed simultaneously, so that all, including the public, medical professionals and regulators alike, can benefit from being able to access a wider information network. Furthermore, co-creation empowers the regulators and policymakers to gain vital insight as much as it encourages stakeholders to gain ownership of the process and its effectiveness. After all, regulation should safeguard and protect individuals, and serve to encourage R&D and not unduly hinder innovation. Indeed, policy, governance and regulatory interventions must not only consider the potential downside of a particular application of an AI system within a given scenario but also the downside of not deploying this system now and potentially in a revised version in the future.

However, as demonstrated in the Round Table Meetings, this balance is delicate and cooperation between Member States, as well as clear guidance at an EU level is needed. Importantly, it has been pointed out that the definition of AI itself and the potential grouping of a range of technologies under one umbrella might pose a challenge to policy and regulatory interventions, and in turn complicate their application.

Moreover, distinctions between the various application of AI systems in the healthcare sector will become crucial for policy as there is arguably a difference between those used in an infrastructural setting (e.g. to sort paperwork and free-up clinicians’ time) and those used in a direct clinical setting. Regulation and policy will need to account for this accordingly.

Finally, it has clearly transpired that shared practices, guidance and benchmarks to inform and implement policy will be needed on a pan-European level. This may be through a lighthouse centre focussed on the healthcare sector, through world reference testing and experimentation sites, through a specialised Digital Innovation Hub network, or new governance mechanisms and agencies.