UK Market Access for AI – 2024 milestones and what’s next
Welcome to the final article in the Decisive Dialogue series on artificial intelligence (AI). 2024 was an impactful year for manufacturers employing AI in their evidence generation or developing AI as a medical device (AIaMDs) in the UK. This article will summarise key events of 2024 from the UK AI market access space and highlight interesting milestones for 2025.
2024 – What happened?
MHRA response to pro-innovation white paper
The 2023 government white paper entitled “A pro-innovation approach to AI regulation”[1] outlined a trans-sector framework for driving AI innovation in the UK. The key principles are (1) safety, security and robustness; (2) appropriate transparency and explainability; (3) fairness; (4) accountability and governance; and (5) contestability and redress.[1] The Medicines and Healthcare Products Regulatory Agency (MHRA) responded to the white paper in April 2024, reflecting on how it plans to adopt this pro-innovation framework.[2] Key points impacting market access include:
· The medical device (MD) risk-based classification system will be reformed. This is anticipated to result in AI products being reclassified to higher-risk MDs requiring greater scrutiny by regulatory assessors.[2]
· To strengthen accountability and governance of AIaMD products, regulations will be updated to clarify and reinforce the obligations of manufacturers, conformity assessment bodies and regulators.[2]
· Predetermined Change Control Plans (PCCPs) will become part of core regulations, initially on a non-mandatory basis.[2] PCCPs are dossiers drafted by manufacturers outlining planned modifications to an MD, such as changes to the model or algorithm, that are expected to have a significant downstream effect on the MD and, therefore, require pre-market regulatory oversight.[3] PCCPs aim to enhance accountability for products over the entire lifecycle of AIaMDs.[3]
These updates align with the MHRA’s Change Programme, which aims to introduce a regulatory framework specifically for AIaMDs and software as a medical device (SaMD) products. The Change Programme launched in July 2024 and is an ongoing effort to supplement MD regulations with more tailored guidance for AIaMDs and SaMDs moving forward.[4]
AI Airlock [5]
In October 2024, the MHRA piloted “AI Airlock”, a regulatory sandbox programme that aims to identify challenges and uncertainties experienced by AIaMD manufacturers in a collaborative way. Four to six currently unknown applicants/ AIaMD products were selected to participate in the first wave.
NICE position statement [6]
In August 2024, NICE published a position statement on using AI for evidence generation and reporting. Detailed methodological and process considerations are to follow; however, the statement offers guidance to manufacturers regarding best practices for market access involving AI.
Key points include:
· Considering the potential risks associated with AI use, it should only be used if there is a demonstrable benefit. Thus, manufacturers need to carefully evaluate whether the use of AI is justifiable and outweighs risks such as lack of transparency, inaccessibility to non-experts and reduced human oversight.
· AI should support and not replace decision-making. It is advised that careful technical validation of any AI applications be conducted, and the results of such validation presented.
· NICE recommends adhering to risk minimisation guidelines, including Cochrane [7], PALISADE [8], TRIPOD+AI [9] and the Algorithmic Transparency Reporting Standard [10]. In terms of reporting and compliance, this recommendation is a significant signpost for manufacturers to understand data quality requirements.
· Manufacturers must submit evidence of how risks, especially for cybersecurity, were addressed. This will require careful documentation of the risk mitigation measures employed throughout the market access process.
2025 – What to look out for
Pro-innovation regulation [2]
As part of the Change Programme, the MHRA announced the release of new, AIaMD-specific guidance for spring 2025 to ensure regulatory processes align with the principles outlined in the government’s pro-innovation white paper. This includes clear guidance on cyber security and detailed supplementary MD guidance on how human factors should be applied.
AI Airlock – Outputs and next steps [5]
AI Airlock terminates in Q1 2025, after which the participating manufacturers and all other parties involved (MHRA, the NHS, UK Approved Bodies and other regulatory agencies) will each report on lessons learnt, challenges, and risks identified. Outputs will inform future sandbox programmes in the short-term, and policy changes in the long-term.
Conclusion
AI is a hot topic for UK regulators looking to introduce clear quality standards and guidance to ensure that only safe and effective products come to market. For manufacturers, it will be critical to stay up to date with new guidance and regulations, as these could significantly impact product development processes, market access and the acceptability of evidence.
Written by Sophia Naegeli
Decisive Dialogue 6th December 2024
If you need help understanding the implications of new regulations to your AIaMDs, please get in touch with us via: enquiries@decisiveconsulting.co.uk or lance.richard@decisiveconsulting.co.uk
Bibliography
1. Secretary of State for Science, Innovation and Technology (2023). A pro-innovation approach to AI regulation [online]. Link: https://www.gov.uk/government/publications/ai-regulation-a-pro-innovation-approach/white-paper. Last accessed: November 27th 2024.
2. MHRA (2024). Impact of AI on the regulation of medical products [online]. Link: https://assets.publishing.service.gov.uk/media/662fce1e9e82181baa98a988/MHRA_Impact-of-AI-on-the-regulation-of-medical-products.pdf. Last accessed: November 27th 2024.
3. MHRA (2023). Predetermined Change Control Plans for Machine Learning-Enabled Medical Devices: Guiding Principles [online]. Link: https://www.gov.uk/government/publications/predetermined-change-control-plans-for-machine-learning-enabled-medical-devices-guiding-principles/predetermined-change-control-plans-for-machine-learning-enabled-medical-devices-guiding-principles. Last accessed: November 27th 2024.
4. MHRA (2023). Software and AI as a Medical Device Change Programme – Roadmap [online]. Link: https://www.gov.uk/government/publications/software-and-ai-as-a-medical-device-change-programme/software-and-ai-as-a-medical-device-change-programme-roadmap#work-packages. Last accessed: December 3rd 2024.
5. MHRA (2024). AI Airlock: the regulatory sandbox for AIaMD [online]. Link: https://www.gov.uk/government/collections/ai-airlock-the-regulatory-sandbox-for-aiamd. Last accessed: December 3rd 2024.
6. NICE (2024). Use of AI in evidence generation: NICE position statement [online]. Link: https://www.nice.org.uk/about/what-we-do/our-research-work/use-of-ai-in-evidence-generation--nice-position-statement. Last accessed: November 27th 2024.
7. Cochrane (2024). Cochrane announces new policy on AI generated content [online]. Link: https://futurecochrane.org/newnews/cochrane-announces-new-policy-on-ai-generated-content. Last accessed: December 5th 2024.
8. Padula WV, Kreif N, et al (2022). Machine Learning Methods in Health Economics and Outcomes Research-The PALISADE Checklist: A Good Practices Report of an ISPOR Task Force. Value Health. Jul;25(7):1063-1080. doi: 10.1016/j.jval.2022.03.022. PMID: 35779937.
9. Collins G S, Moons K, et al (2024). TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ 385:e078378 doi:10.1136/bmj-2023-078378
10. Department for Science, Innovation & Technology (2023). Algorithmic Transparency Recording Standard - Guidance for Public Sector Bodies [online]. Link: https://www.gov.uk/government/publications/guidance-for-organisations-using-the-algorithmic-transparency-recording-standard/algorithmic-transparency-recording-standard-guidance-for-public-sector-bodies. Last accessed: December 4th 2024.