THE UK’s medicines regulator has predicted that artificial intelligence (AI) will be one of the “defining medical technologies” of the next few years in their new strategy.
The Medicines and Healthcare products Regulatory Agency (MHRA) will lay out its intentions to set a global standard for safe and agile regulation of MedTech and AI in healthcare in its new strategy to 2030, with the NHS aiming to become the world’s most AI-enabled health systems globally.
The proposal, to be released on Wednesday at 10am, will cover AI-enabled products, medicines development, the use of AI in diagnostics and treatment, as well as safety monitoring and the agency’s own regulatory processes.
It will also outline the agency’s plan to establish a Centre for AI and Digital Health and publish an AI Medicines Development Action Plan, combining support for innovation with stronger safety oversight.
The report, which defines the MHRA’s eight-pronged approach to achieving ‘better outcomes’, will highlight AI and its use in healthcare across all four UK nations as a key focus point.
Acknowledging that AI medical devices present distinct challenges because performance can change over time, between healthcare settings and across patient populations, the MHRA plans to enact continuous monitoring of real-world performance, with clearer accountability across developers, healthcare providers and professionals.
While foundation models, trained on broad datasets and adaptable to multiple uses, add further complexity, its Expanded Predetermined Change Control Plans will enable pre-agreed model updates, while proposed Master Files could allow developers to share essential safety and performance information about underlying foundation models.
The agency will also share plans for a new Centre for AI and Digital Health which will bring together specialist expertise to support regulation across medicines and medical devices, making requirements clearer for developers and healthcare partners.
Supporting Medicines Development
A brand-new AI Medicines Development Action Plan, to be released in 2027, will set out a coordinated UK-wide approach covering discovery, pre-clinical modelling, clinical trial design, manufacturing and post-market monitoring.
In the report, the agency will also share how it is already incorporating AI considerations into chemistry, manufacturing and controls frameworks, piloting synthetic data in clinical research and exploring AI predictions of how medicines behave in the body and their potential toxicity.
Working with the Regulatory Innovation Office, its ‘Beyond ADMET: AI for Medicines Safety’ sandbox is examining responsible use of AI-generated evidence.
Further proposals will also include statutory mechanisms for sandboxes, subject to legislative reform, alongside guidance on digital twins and support for hybrid trials using synthetic data.
Through its AI for Regulatory Insight, Safety and Efficiency (ARISE) programme, the MHRA also plans to use AI for application triage, evidence screening and identifying missing information.
An AI-enabled Clinical Trial Protocol Checker will aim to help sponsors identify common grounds for non-acceptance before submission, with the tools supporting human expertise.
Large volumes of NHS date will also be put through AI to detect emerging safety signals and support earlier intervention, stronger horizon scanning will identify new technologies sooner, and evidence-based communications will address misleading information on social media and AI platforms.
The MRHA hope that international cooperation, including through the Health AI Global Regulatory Network, will support shared safety signals and standards, with closer engagement on AI-enabled healthcare also planned with India and the Middle East and North Africa.
Finally, the agency will expand its Leeds Digital Hub, strengthen digital skills and explore AI automation in HR services. Across these plans, the strategy emphasises public trust, transparency and patient safety.
Read the full report here at 10am.
Featured image: ipopba on AdobeStock
Each article is made available under the terms of the Creative Commons Attribution-Non Commercial 4.0 License.
