MHRA releases new framework 

Rates of depression, anxiety, and insomnia continue to rise globally1 and traditional services are struggling to meet demand. A statistical breakdown in the US shows that 19.1% of adults had an anxiety disorder and 14.5 million adults experienced a major depressive episode in the past year.2 At the same time, around two-thirds of adults with mental health issues were unable to access treatment.3 To bridge the divide between demand and treatment availability, digital tools like apps, wearables, and AI-powered platforms are emerging as vital tools to close the treatment gap by offering more accessible and tailored support.

It’s not surprising therefore that the global market for emerging mental health devices and platforms was valued at $3.22 billion in 2023 and is projected to reach $17.70 billion by 2031.4 Digital health technologies have the potential to transform the landscape of mental and physical healthcare by making support more accessible, personalized, and data-driven. These tools can in fact help manage a range of issues such as asthma with smart inhalers, diabetes with continuous glucose monitors (CGMs), blood pressure via hypertension monitoring platforms, heart conditions via wearable cardiac monitors, and post-surgery recovery via trackers and remote physiotherapy, to name just a few. In addition to delivery formats, they can also be available as direct-to-consumer products or used with a referral or supervision from healthcare or educational professionals.

The rapid growth and complexity of these technologies have, however, raised urgent questions about regulation and safety, especially when it comes to their integration of large language models (LLMs) and AI that go beyond simple tracking and monitoring of physical data by providing actual diagnostic or treatment purposes.

Key criteria for DMHT classification: Assessing intended purpose and functional impact
In this context, the UK Medicines and Healthcare products Regulatory Agency (MHRA) has provided a new critical piece of guidance5 that will help inform, specifically, digital mental health technology (DMHT) manufacturers that want to put their devices on the UK, market. The guidance is a key output from the Wellcome Trust project, part of a £1.8 million investment by the UK government to explore the regulation of digital mental health tools. The initiative reflects the increasing need for a structured regulatory framework that balances innovation and safety. It also provided key intelligence for any manufacturer wishing to place, or indeed to keep, their devices on the UK market.

The MHRA’s newly released framework offers long-awaited clarity by outlining when DMHTs qualify as medical devices based on their intended purpose and functional impact. This guidance also addresses the emerging role of AI and generative models in mental health applications, highlighting the balance between technological advancement and patient protection.

Specifically, the MHRA provides a framework for determining whether a DMHT qualifies as a medical device. It is based on various aspects of the tool’s design, functionality, and intended use. The first step in qualification is in fact assessing what the tool claims it is designed to do. Software that explicitly states that it diagnoses, treats, prevents, or monitors a medical condition would be more likely to be subject to medical device regulations. In addition to this, a tool’s functionality now plays a decisive role. A DMHT with a medical purpose but low functional impact could well be excluded from medical device regulation compliance as it does not provide a clinical effect or influence patient care decisions.

If a product has been flagged as a potential medical device, manufacturers may be asked to complete a Device Characterization Form. It may also be compiled proactively. In both these cases the MHRA will analyze information such as websites or screenshots and may request additional information and delve into intended purpose and functionality definitions. Purpose and functionality will be evaluated taking into account labeling, instructions for use (IFU), and promotional materials as well as technical documentation.

Basic tools that add up a series of numbers as scores to provide a result (e.g., a PHQ-9 calculator) are typically classified as low functionality as a result of these considerations. Devices that incorporate user sign-in via a generative-AI chatbot that responds adaptively based on the user input, on the other hand, would be considered high functionality. Tools where a patient answers a large number of AI-tailored questions, for example, and that show the likelihood of mental health conditions as a result would be considered high functionality. Other examples of low functionality tools that do not fulfill a clinical role are transcription tools used by primary care physicians during appointments, so-called ambient scribes. If the consultation is recorded and summarized in writing to be recorded in an electronic record, however, functionality is high; the clinician may not have time to check every summary, putting the patient at risk.

Navigating international AI and SaMD Regulations: US, UK, And EU frameworks
The FDA has also provided guidance such as its Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan6 but, unlike the MHRA framework, it suggests that devices will need to be evaluated on a case-by-case basis. For US medical device manufacturers with a footprint in the UK, it is critical to become familiar with this guidance and also with the MHRA’s Software and AI Change Programme Roadmap,7 which provides a program of work to ensure regulatory requirements for software and AI are clear and patients are protected. Understanding the requirements in the UK will also provide a solid ground for entry or permanence on the EU market, where legislation that covers AI-based medical devices such as the EU’s AI Act, effective August 1, 2024, introduces a comprehensive regulatory framework for AI.

Medical device manufacturers developing AI-powered DMHTs should thus sit up and take notice of the new MHRA guidelines as they present a blueprint for international regulators’ efforts to provide clear pathways for innovation to thrive while also preserving patient safety. Keeping up to date with the regulatory landscape around AI and DMHTs will prove critical to building trust with regulators, patients, and users and creating longevity for devices on their markets. Med Device Online