The Medicines and Healthcare products Regulatory Agency (MHRA) recently published a roadmap for “Software and Artificial Intelligence as a Transforming Program for Medical Devices”. This builds on broader reforms in the pipeline, explained Claire Burroughs, a regulation and compliance expert at Walker Morris.
The future regulation of medical devices in the UK – where are we now?
He advised the MHRA in 2021 on the new post-Brexit regulatory framework for medical devices. The government’s response was published in June this year. We explained in our last briefing that the MHRA has now extended the implementation of the Future Medical Device Regulations by 12 months. It aims to enter into force by July 2024.
What about software and artificial intelligence as a medical device?
Chapter 10 of the MHRA Consultation deals with Software as a Medical Device (‘SaMD’) including Artificial Intelligence as a Medical Device (‘AIaMD’). SaMD is both standalone software and software included in wider devices including AIaMD.
In its response to the advisory, the government said that software and artificial intelligence as a medical device have grown in market share, public health importance and complexity in recent years; The current regulations contain a few provisions that are specifically intended to regulate them. However, the government has made it clear that the majority of change needed in this area is likely to be in the form of guidance rather than legislation. For example, clarifying what medical device requirements mean in the context of software and artificial intelligence.
MHRA announced the program and artificial intelligence as a program to transform medical devices last year. This program of work builds on the broader reforms detailed in the consultation. It will “deliver bold steps to provide a regulatory framework that provides a high degree of protection for patients and the public, but also ensures that the UK is recognized globally as the home of responsible innovation for medical device programs with an eye to the global market.”
The program consists of 11 work packages across two work streams. The former contains eight workpacks for major fixes across the SaMD lifecycle. These include: classification; Pre-market requirements Market mail and cyber secure medical devices. The second contains three work packages and takes into account the challenges that AIaMD can pose over classically programmed programs. This is: interpretation. and adaptation.
The roadmap for each action package describes: the problem the package seeks to address; objectives that detail this problem; and specific outputs that meet the ambition. Workpacks 6, 7, and 8 do not have separate outputs assigned to them. Its content has been assimilated into the broader deliverables identified.
Crucially, the MHRA says it is sensitive to the fact that regulatory innovation that deviates from the international consensus could create an additional burden on the marketplace. It intends to drive the international consensus forward, and work with others to accelerate regulatory innovation and harmonization in the field.
MHRA also works closely with the British Standards Institution (BSI) to ensure that a wide range of standards are in place to help manufacturers meet regulatory requirements.
what happened after that?
The MHRA will announce plans to engage the patient/public and industry in support of the program of work; Most outputs will be released first as drafts for wider input and comment before being published.
Deliverables will be posted in slides. First drafts of the first tranche are due to be published before the end of the year. The first tranche consists of: organizational guidance on crafting an intended purpose in a SaMD context; review of the reverse incident signal detection process for the SaMD; regulatory guidance on machine learning good practices for mapping medical device development; Best Practice Guidelines on AIaMD for Everyone; and Guidelines for Adaptation and Change Management in AIaMD.
We will continue to monitor and report on developments.