What does an ‘enabling’ regulatory environment look like for AI solutions in health?

Charlotte Stix Headshot
Round Table Co-Chair, Charlotte Stix, is an expert in European policy on artificial intelligence (AI). Read her thoughts on what factors need to be considered when regulating AI for use in healthcare settings.

“First and foremost, an enabling regulatory environment must take into consideration all of the varying factors within the ecosystem and their responsibilities.”

Oftentimes it may be unclear where responsibility lies, especially with regard to liability and risk assessment. There is an urgency to combat fragmentation and the ensuing uncertainty both at the granular level (i.e., between developers, national bodies and healthcare professionals [HCPs]) as well as on a more macro-level. The latter is especially key to ensure that the European Union (EU) creates proportionate regulation within an enabling environment. This means, for example, to encourage a pan-European infrastructure that sets a common direction while empowering coordination and cooperation between national regulatory agencies. Cooperation and coordination can also solve practical issues that may accompany regulation, caused by a lack of knowledge sharing.

“Second, it is clear that while vast societal benefit can be derived from health data, it is also highly sensitive. An enabling regulatory environment must strike a balance between this.”

We already see the EU’s ambition to establish common European Health Data Space to support research, as well as a more seamless flow of information between relevant agencies (e.g., with electronic health records). At the same time – in order to account for the nuances in healthcare, as well as for the changing capabilities and application areas of AI systems – risk assessments, testing and experimentation facilities should supplement and support the development of a regulatory environment. An EU-wide network of such infrastructures can ensure that regulation is sufficiently flexible and adaptive to the changing landscape.

Third, education is key. This encompasses knowledge obtained when handling AI systems in healthcare in different jurisdictions, as well as ensuring that there is a consistent and clear flow of information to inform investors and start-ups about regulatory developments within the EU.

“Finally, an enabling regulatory environment must consider the cost (societally and otherwise) of not using certain AI applications in a healthcare setting.”