Regulating AI in Healthcare: National Commission sets out UK framework

The National Commission into the Regulation of AI in Healthcare has published recommendations expected to underpin the future UK regulation of AI in healthcare. The report seeks to support the safe and effective adoption of AI technologies across the healthcare system while ensuring safeguards are in place to maintain confidence and benefit in their use.

The recommendations reflect the increasing role of AI in healthcare delivery, diagnostics, medical research and clinical decision-making and form part of a broader discussion regarding how existing regulatory regimes need to adapt to rapidly evolving technology, including AI. 

The report was compiled by a group of leading regulatory experts, clinicians and industry figures, with working groups seeking views from across the sector. The recommendations contain the expected shift in focus from relying heavily on just the initial regulatory approval, to ongoing lifecycle monitoring. If enacted, this will place an increased burden on providers and adopters of AI, in terms of collecting and responding to performance data and usage trends. However, it is hoped that will in turn improve safety, build trust among healthcare professionals and patients, and ultimately lead to increased adoption of innovation at scale. 

What are the recommendations? 

The Commission's 44 recommendations are grouped into three areas,:

1. Proportionate lifecycle regulation, ensuring that regulatory approaches are tailored to each AI tool’s distinctive characteristics and risk factors, thereby facilitating safe and effective use (Recommendations 1-23). For example:

• Recommendation 6 (Predetermined Change Control Plans) to introduce a change management approach for AI-enabled devices whereby such plans provide for adaptive AI-enabled devices and guardrails for the scope of any planned changes (an approach that has been implemented by the FDA in the US). 

• Recommendation 18 (Reporting and information sharing across the lifecycle) to ensure clear and, where appropriate, standardised mechanisms for reporting and sharing information, thereby enabling a better understanding of ongoing performance and safety as well as improved awareness of reporting routes. 

2. System-wide responsibility & safe management, emphasising that key stakeholders (including manufacturers, healthcare service providers, healthcare professionals, regulators and policymakers) all have a role in ensuring the safe and effective use of AI in healthcare (Recommendations 24-34). For example:

• Recommendation 28 (Contractual allocation of responsibilities) to ensure clear allocation of responsibilities in contracts between manufacturers and healthcare providers, including in respect of the delivery of required risk controls and meeting regulatory requirements. 

• Recommendation 30 (National leadership for AI readiness and adoption) for the Department of Health and Social Care and the devolved nations’ health departments to provide guidance which defines best practice in AI readiness and associated expectations for organisations, as well as a national learning function for collective learning.

• Recommendation 33 (Provision of training by healthcare provides) for providers to ensure that staff are well-equipped to use AI technologies in their practice. 

3. Trust, transparency and predictability, ensuring that public confidence is achieved through transparency and responsible development, whilst the benefits of innovative AI in healthcare can be realised (Recommendations 35-44):

• Recommendation 37 (Public communication on safety concerns) for the MHRA to put in place mechanisms ensuring public-safety information is clear, transparent and tailored to the needs of the public.  

• Recommendation 42 (Early regulatory advice and pathway confirmation) for the MHRA to issue predictable engagement mechanisms for developers and manufacturers, including a route providing written feedback at an early stage on areas such as device classification, the path to market and post-market expectations.

Overall, the recommendations recognise that AI presents significant opportunities for improving patient outcomes and healthcare delivery but also raises unique challenges requiring appropriate governance, oversight and adaptability.

Associated Guidance

In parallel with the report’s release, NHS Resolution has issued guidance on scheme coverage and liability issues concerning the use of Artificial Intelligence in the context of clinical practice. The guidance explains how the existing NHS indemnity scheme and liability principles regarding clinical negligence apply to situations where AI is being used in the delivery of healthcare.

The Guidance demonstrates alignment with the Recommendations, including in areas such as:

• Expressing the importance of appropriate human oversight of the AI tool;

• Clarity of roles and responsibilities of key stakeholders including the AI tool’s manufacturer and the person and/or organisation using it;

• Facilitating AI-readiness, including staff training and ensuring robust governance arrangements are in place;

• Continuous monitoring of the AI tool throughout its lifecycle.

Wider regulatory reform

The Recommendations come against a backdrop of wider efforts, including a range of AI sandboxes we discuss here, to update the UK's AI in healthcare regulatory framework.

Additionally, as discussed in our recent article on proposed reforms under the UK Health Bill, the Government has tabled amendments designed to expand the MHRA's information-sharing powers, create a more agile framework for updating medicines and medical device legislation, and support the future development of a medical device licensing regime. These proposals are intended to ensure that the regulatory system can respond more effectively to scientific and technological developments, including the increasing use of AI within healthcare. 

Taken together, the Health Bill reforms and the Recommendations demonstrate a clear direction of travel towards a more modern and adaptable regulatory framework capable of supporting innovation while maintaining appropriate safeguards.

What happens next?

The Recommendations do not themselves introduce any immediate legal or regulatory changes, they will inform future policymaking and regulatory development. In the short term, attention now turns to the government who we expect to publish a cross-government response setting out how it plans to implement the Recommendations. In the medium term, policymakers will likely engage further with industry stakeholders, devolved authorities and health system partners before any substantive reforms are implemented.

Looking ahead

Organisations involved in the development, procurement, deployment or use of AI-enabled healthcare tools should closely monitor regulatory developments and assess their current governance arrangements against the direction of travel identified by the Commission. 

Together, the Commission's report and the Guidance suggest an increasingly mature regulatory environment in which manufacturers will be expected to demonstrate the safety, effectiveness and ongoing governance of their products, while healthcare providers and clinicians will be expected to ensure those products are deployed, monitored and used safely and appropriately. As AI adoption accelerates, issues such as human oversight, organisational governance, post-market monitoring, appropriate contractual terms and insurance arrangements are likely to come under increasing scrutiny.

While significant questions remain regarding the precise shape of any future reforms, the publication of the Commission's recommendations represents another important step in the development of the UK's approach to regulating AI in the healthcare sector. We will continue to track the implementation of the recommendations and any consequential regulatory reforms.

If you would like to discuss further, please contact a member of our Health, Care and Life Sciences Team.

This article was written by Susannah Jury and Issy Peacock.