Transformation Directorate

AI regulation guide: considerations when developing AI products and tools

AI regulation guide.png

A guide from the AI and digital regulations service on the importance of the value proposition, determining a product’s intended use, and planning ahead to navigate the path to market.

Common challenges in AI development for health and care

Everything begins with having a clear, purposeful value proposition; and less is often more.

Determining a realistic expectation of the value a potential product could bring to the health and care system (i.e., the value proposition) and demonstrating it convincingly is fundamental to success. This key message is the basis of everything that follows, and it should provide clarity on what a product is aiming to do (its “intended use”, in regulatory speak), and what benefits it could bring. This clarity is important for developers navigating product development and the regulatory pathway as well as for potential adopters to understand the value a product would bring.

Feasibility is a critical element of your value proposition, meaning whether or not it is possible to generate evidence to support the value claim. Generating robust evidence that could convince NICE and commissioners to support just one value claim requires a large amount of time and resources; an all-too-common mistake is over-promising on the value that a product can bring, and consequently under-delivering on the evidence to support these claims. At NICE, we often observe that one or two value propositions that are well-evidenced is regarded more favourably than several plausible, but nonetheless theoretical, claims. So, when developing a product, you should start by:

  • considering at an early stage whether it is possible to generate evidence to support each value claim
  • thinking about the order in which you want to generate this evidence; maybe you want to evidence all value propositions, but are starting with near-term value and planning for longer term data collection
  • avoiding the temptation of over-selling a product; at the risk of spreading resources too thinly.

See the NHSX guide to good practice for digital and data-driven health technologies, for more guidance on developing a clear and robust value proposition, and the NICE evidence standards framework for digital health technologies for guidance on how to evidence these claims.

Aligning the value proposition and intended use of a product

Developers often overlook the need for value propositions to align with intended use.

So how do you define the ‘intended use’?

A statement of intended use clarifies exactly what your product is used for (i.e., its purpose), and how it should be used. This is critical for establishing the scope of a product and the required regulatory activities for developing it. You will need to define and document the intended use of your product to launch a product on a regulated market such as the health and care system.

The intended use of a product determines when it qualifies as a medical device for the regulatory framework as well as the MHRA classification of a product. Determining whether a product is a medical device, and if so what class, is often more complex than it appears. For instance:

  • a product that periodically collects information about a person’s movements is not a medical device if marketed for fitness and general consumption
  • taking the same product and changing its intended use to tracking movements in a care setting for people with mobility issues might make it a medical device if the intention is for the information to be used for a medical purpose.

The intended use should align with the value proposition claims made to commissioners or to NICE. For example, a product with an intended use to offer diagnostic support that has a human in the loop at many points may deliver efficiency benefits as part of its value proposition; but these would likely be modest compared with a fully automated system. Yet sometimes, when making value claims, developers describe efficiency benefits closer to what would be expected in a fully automated system, which suggests they might not fully understand their product’s regulations and relatedly may have struggled to comply with them. It may also suggest that regulatory checks appropriate to the product’s intended use have not been undertaken, which means the technology could be liable to safety issues.

So, when developing a product, it is critical to ensure that the intended use and value proposition are aligned at an early stage so that you go down the right regulatory pathway for both together. To your benefit, this should prevent wasted resources on inappropriate evidence generation.

The bottom line: planning ahead is key

Developers should think carefully about their path to market, and not underestimate the time and resources required to navigate the regulatory pathway.

We know that the legal, regulatory framework governing the development of new technologies is complex and challenging to navigate. This can be daunting for developers, who understandably want to reach market access as quickly as possible. Yet because of this, developers often have a good feel for the regulatory requirements for the current state of a product, but do not have a plan for navigating the pathway if and when they make significant changes to their product. This can be a particular issue for AI, because of its ability to change and evolve much more quickly than other technologies.

Imagine you are developing a product to improve hospital system efficiency and reduce admin burden, but in the long run want to develop this into a predictive tool for triaging patients. The latter would plausibly affect treatment outcomes for patients, and in such a case, the product may evolve from a non-medical device into a medical device. So, when developing a go-to-market strategy, you should think all the way through from regulations that apply to a product now, to regulations that may apply as a product is developed further. This will allow you to plan for possible regulatory requirements and plan product development and evidence generation around this.

Planning ahead in this way will ensure you aren’t blindsided by further regulatory requirements as you embark on your product development strategy.


Further information

About the AI and digital regulations service

The MHRA website has a range of resources and guidance can be found on the regulation of medical devices.

The Digital Technology Assessment Criteria (DTAC) helps with assessing suppliers and gives developers what is expected for entry to the NHS.

Get details on what CQC registration is, and who needs to register.

The Health Research Authority’s website, outlines what approvals are required for health research and how to obtain them.


Rebecca Boffa, Jeanette Kusel, Russell Pearson, Moritz Flockenhaus, Omar Moreea, Carly Wheeler, Toni Gasse, and Clíodhna Ní Ghuidhir1, on behalf of the MAAS working group, and in collaboration with Emma Hughes, on behalf of the AI Award team at the Accelerated Access Collaborative.

For more information on the AI and digital regulations service, contact Clíodhna Ní Ghuidhir (corresponding author) at

AI regulation guide.png