Regulators Adapt to AI Revolution in Healthcare: MHRA Leads with New Guidance and Sandbox Initiative
Key Insights
The UK's MHRA has published comprehensive guidance on regulating AI in healthcare while exploring ways to leverage the technology to improve its own regulatory efficiency.
A new regulatory sandbox called "AI-Airlock (search)" will launch in spring 2024, creating a controlled environment for testing AI as Medical Device (AIaMD) products and identifying potential regulatory challenges.
Regulatory bodies worldwide are developing frameworks to balance innovation with safety concerns, with the EU finalizing its AI Act and the FDA implementing pre-authorized software change processes for AI/ML systems.
Regulatory agencies worldwide are rapidly adapting their frameworks to address the accelerating integration of artificial intelligence in healthcare and pharmaceutical development, with the UK's Medicines and Healthcare products Regulatory Agency (MHRA) taking a leading position through newly published guidance and innovative regulatory approaches.
The MHRA recently outlined its strategy for both regulating AI-powered medical products and utilizing the technology to enhance its own regulatory processes, responding to the UK government's ambition to become an "AI superpower" as outlined in a recent white paper.
UK's Three-Pronged Approach to AI Regulation
The MHRA has identified three critical areas where it must engage with AI technology: direct regulation of AI used for medical purposes, adoption of AI to improve its own regulatory services, and developing sufficient understanding of how regulated companies are implementing AI in their operations.
"AI offers us the opportunity to improve the efficiency of the services we provide across all our regulatory functions from regulatory science, through enabling safe access for medicines and medical devices, to post-market surveillance and enforcement," said Laura Squire, chief quality and access officer at the MHRA.
The agency's approach aligns with five principles outlined in the UK government's white paper: safety and security, appropriate transparency, fairness, accountability, and contestability.
AI-Airlock: A Regulatory Sandbox for Innovation
Central to the MHRA's strategy is the launch of "AI-Airlock (search)," a regulatory sandbox scheduled to begin in spring 2024. This controlled testing environment will allow developers to trial AI as a Medical Device (search) (AIaMD) products under regulatory supervision before wider implementation.
The sandbox aims to identify regulatory challenges specific to AI-powered medical devices and develop solutions in collaboration with stakeholders across healthcare regulation, governance, and assurance organizations.
Additionally, the MHRA plans to release detailed guidance on cybersecurity and human factors for AIaMD products by spring 2025, addressing two critical aspects of AI implementation in healthcare.
Internal AI Applications to Enhance Regulatory Efficiency
The MHRA is also exploring how AI can streamline its own operations, particularly in the initial assessment of marketing authorization applications. The agency is investigating supervised machine learning to assist human assessors in document review, potentially accelerating the approval process while maintaining rigorous standards.
Other internal applications include using generative AI and large language models to derive actionable insights from real-world data and enhance vigilance systems for monitoring product safety.
Global Regulatory Landscape Evolving
The UK's approach comes amid broader international efforts to regulate AI in healthcare. The European Union has published the final text for its AI Act, which will regulate AI systems across multiple industries, including medical devices. The Act classifies certain healthcare AI applications as "high-risk" due to their potential impact on fundamental rights.
In the United States, the FDA has compiled a publicly available list of AI-enabled tools with marketing clearance and has introduced processes for pre-authorized software changes, allowing certain AI systems to be adaptively re-trained on new data without requiring additional regulatory assessment.
Balancing Innovation and Safety
A key challenge for regulators is distinguishing between digital health technologies and digital medical devices. The International Medical Device Regulators Forum (IMDRF) defines Software as a Medical Device (search) (SaMD) as "software intended to be used for one or more medical purposes that perform these purposes without being part of a hardware medical device."
The MHRA is developing guidance to help clearly identify SaMDs and differentiate them from wellbeing and lifestyle software products, IVD software, and companion diagnostics.
Pharmaceutical Applications and Challenges
The pharmaceutical industry faces unique challenges with AI implementation. Advanced therapeutics like MSD and Moderna's potential cancer vaccine (search), which uses AI to target antigens specific to each patient, represent a new class of treatments that evolve based on the data they process.
This adaptability, while offering tremendous potential for personalized medicine, presents regulatory challenges for agencies accustomed to evaluating fixed, traditional pharmaceutical products.
Collaborative Path Forward
As AI continues to transform healthcare delivery and pharmaceutical development, collaboration between industry, regulators, and healthcare providers will be essential. The MHRA's approach emphasizes stakeholder engagement, including efforts since 2020 to work with the pharmaceutical industry on AI applications for vigilance purposes.
The agency acknowledges potential challenges ahead, including the increased pace of new medicines developed through AI, changing clinical trial designs, and AI's role as an enabler of personalized medicine.
For developers and manufacturers navigating this evolving landscape, engaging with regulatory experts and embracing multidisciplinary approaches will be crucial to ensure compliance while advancing innovation that can transform patient care.