Digital health is one of the fastest growing and most dynamic areas within the UK healthcare and life sciences ecosystem.

Innovators are benefitting significantly from developments in AI and the increased adoption of digital and cloud-based technologies, both by over-pressed healthcare organisations and consumers.

Whilst there is plenty of opportunity for exciting innovation and further growth, in a recent publication the Association of British HealthTech Industries (ABHI, the industry body for health tech) highlighted some of the challenges digital health companies are facing in the UK. These include regulatory blockers, challenging innovation and procurement pathways within the NHS, and the need for long-term strategic approach and investment in the sector. These issues, and the corresponding action points set out by the ABHI, will be for the next government to address as part of its business. In the meantime, there are updates from the MHRA which are relevant for digital health companies and their investors.

Use of AI models in software classed as medical devices

Earlier this year, the UK government published its non-statutory framework on AI regulation which included a requirement for different regulators to publish specific, sector relevant AI guidance, and the Medicines and Healthcare products Regulatory Agency (MHRA), has recently updated its AI guidance in response.

For general context, the MHRA has long ago recognised that the current rules relating to software as a medical device are due for an update. Along with wider reforms of the medical device regulations, it is currently undertaking an extensive change programme to modernise the regulation of software products and apps used as medical devices. This programme includes AI-specific workstreams. It has also collaborated internationally with other regulators to create guidelines on good machine learning practices and in connection with change control for machine leaning models in medical devices.

The MHRA shapes its AI guidance

Within the recently published guidance, the MHRA sets out how it plans to meet the five principles of regulatory use of AI set out in the government white paper. The five principles are:

• Safety, security and robustness
• Transparency and explainability
• Fairness
• Accountability and governance
• Contestability and redress

When it comes to transparency and explainability, relevant guidance already exists, which will be supplemented in due course with further guidance on the user interface aspects of AI solutions. New guidance is also awaited on cyber security.

To ensure fairness of the AI models, developers are urged to comply with current international standards such as the ISO standard on information technology, artificial intelligence, bias in AI systems, and AI-aided decision making (ISO/IEC TR 24027:2021) as well as the relevant International Medical Device Regulators Forum guidance.

Some of the principles will be further addressed in the updated medical devices regulations. For example, the MHRA intends to introduce pre-determined change control plans for AI models, on a non-mandatory basis to start with, to improve accountability and transparency on the datasets used to train them.

Interestingly, the guidance also states that, following the updates to the medical device regulations, the MHRA expect many AI software products to be moved up to a higher risk category, leading to increased transparency obligations for the relevant applications.

Finally, the EU AI Act may be relevant for digital health companies targeting Europe. This adds an extra layer of complexity, as it imposes further obligations in addition to the applicable medical device requirements. The Act was approved in March this year and most of its provisions become effective after a two-year transitional period.

MHRA regulatory sandbox for AI

The MHRA has further announced a regulatory sandbox specific to AI, known as the Airlock, which will soon accept applications. This will bring together regulators, the industry, as well as key players such as the NHS to address the unique challenges posed by AI for the regulation of medical devices.

Digital mental health technologies

Consistent with increased attention to mental health issues in public discourse and media, the number of digital mental health solutions such as mindfulness and therapy apps has also grown. The MHRA is carrying out a project focussing specifically on digital mental health technologies and has published an initial report on user experiences and expectations of these apps. The research report indicates that these technologies are seen as a potentially useful element of a wider treatment approach. Views were generally in favour of regulation and/or guidance for providers of these technologies, with ease of use, data security and accuracy of advice and diagnosis highlighted as particular concerns by users.

Following the publication of this research, the MHRA will now go on to release further information on its proposed approach to digital mental health technologies.

For further guidance and support please contact our healthcare and life sciences experts.

Related News, Insights & Events