MHRA ‘sandbox’ to pilot AI regulation will go live soon

Gerd Altmann

The UK’s Medicines and Healthcare products Regulatory Agency (MHRA) has said it will launch a pilot programme in the coming weeks to see how it can evolve the regulation of artificial intelligence as a medical device (AIaMD) technology.

The AI-Airlock regulatory sandbox will bring together the NHS, regulatory authorities, and approved bodies – organisations that have been designated by the MHRA to assess manufacturers and their medical devices – to “start to identify and address the novel regulatory challenges for AIaMD.”

It was first announced last year and will provide a regulator-monitored virtual area for developers to generate robust evidence for their advanced technologies, according to the agency, which says it will be ready sometime this spring.

The timeline is laid out in a new MHRA strategy document that aims to update and reform the regulation of AI used for medical purposes, part of an ongoing update to its framework for medical devices.

According to the MHRA’s chief quality and access officer, Dr Laura Squire, there is a need for “risk proportionate regulation of AIaMD […] which takes into account the risks of these products without stifling the potential they have to transform healthcare.”

The reforms will focus on five key principles: safety, security, and robustness; appropriate transparency and explainability; fairness, accountability, and governance; contestability; and redress, according to the document.

“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 Squire.

“Increasingly, we expect AI to feature in how those we regulate undertake their activities and generate evidence and we therefore need to ensure we understand the impact of that in order to continue to regulate effectively.”

AI-Airlock will answer the challenges that can be associated with testing AIaMD using traditional trial techniques and may mean that patients could benefit from faster access to developing technologies, such as improved diagnostics or precision medicine, according to the MHRA.

Along with AI-Airlock, the MHRA is also working on two new guidance documents, called Good machine learning practice for medical device deployment and Best practice AIaMD development and deployment, adding to several other guidance documents published by the regulator on software as a medical device (SaMD) technologies.

It is also exploring the use of AI to perform an initial assessment of documents submitted as part of the marketing applications, to analyse real-world data to understand how the use of medical products can affect clinical outcomes, and to protect consumers from fraudulent products.

Image by Gerd Altmann from Pixabay