Carly Wheeler, Policy Manager (Data and AI)
Our Policy Manager for Data and AI, Carly Wheeler, writes about her role in developing a multi-agency advisory service for developers and adopters of artificial intelligence (AI) and data-driven tech. In this blog post, she outlines four key things developers and adopters have told us they want from the service, and how you can get involved.
Funded by the NHS AI Lab, the HRA is working with the Care Quality Commission (CQC), Medicines and Healthcare products Regulatory Agency (MHRA) and the National Institute for Health and Care Excellence (NICE) to develop a multi-agency advisory service (MAAS) for developers and adopters of AI and data-driven technologies. My project team colleagues and I are determined that the MAAS will be built around genuine user needs. So, over the last few months, we’ve been busy conducting in-depth user research interviews. These findings reflect the initial round of ‘discovery’ research, where we spoke to 10 developers, nine adopters and 15 representative organisations such as industry bodies (and it's now increasing each week!) These interviews have given us a fantastic insight into opportunities for the MAAS to tackle the most pressing challenges developers and adopters of AI face.
Clear and direct advice
Many interviewees told us that they found it difficult to know which regulatory processes they must comply with and which guidance it is helpful to consider. Differentiating between these, and explaining why, would be of huge benefit. One user summed it up by saying:
'innovation-friendly regulation means: do this and we’ll regulate you'.
The MAAS will provide clear, actionable information, written in accessible and jargon-free language. Where there is a regulatory requirement, we'll say why this is necessary from a legal and/or ethical perspective. But where there is ambiguity or a judgement-call to be made on a regulatory area that cannot be distilled into something pithy, we'll explain why and give advice on how best to navigate it. As new regulations evolve, we'll make sure these changes are explained to users in a timely way.
Joining up the multiple regulators
There is currently no one-stop-shop where adopters and developers can go to for advice. One interviewee described how they:
'have to go from agency to agency to find the information needed.'
Both developers and adopters perceived a lack of clarity on the roles of the multiple regulatory bodies, which regulations sit with which body, and where along the product development and deployment pathway these regulations apply.
As a cross-regulatory service, the MAAS will provide a joined-up overview of the regulatory and market access pathway, which includes health technology assessment, mapping legislation to the relevant organisation, and indicating clearly to users what they need to do when. This will be critical in helping users get their regulatory requirements right first time, reducing unnecessary delay.
As regulatory bodies, we'll also be thinking about how we work together, aligning our requirements where possible. This should avoid duplicative activities and reduce the burden of compliance. We're actively exploring this, and hope to share more in the coming months.
Linking in with the wider AI ecosystem
Developers and adopters told us that they do not just go to national regulatory bodies for advice on regulation or market access. Their first port of call is often their own personal networks, local research networks in NHS trusts or academia, industry support organisations and innovation support hubs. This signalled to us this an opportunity to increase our reach and deepen our relationships with the wider innovation support ecosystem.
In developing the MAAS, we'll be engaging with partners across the ecosystem. We must make sure that they too are learning about the regulatory pathway as it evolves so that their advice to key users remains up to date. It’s vital that this engagement is two-way. They can consult with us on complex regulatory questions and we can learn from them; learning what challenges our users experience on the ground and refining our advice in response to user needs.
Building an AI regulatory system with trust at its core
Through the MAAS, we hope to build trust in how AI in health and social care is regulated. This theme came out strongly during our research.
Developers want regulatory bodies to demonstrate that we're supportive and want to enable innovation. We know we must be also be transparent about processes, and work to minimise friction in the path to market access.
Adopters want regulatory bodies to provide confidence in the products developed through robust appraisals of safety, effectiveness, and fairness. Users of services need the same confidence in products as well as on how their data is used.
Achieving these four key requirements is no small task. But we’re confident that the multi-agency aspect of this collaboration – bringing strengths together across more than just the MAAS partners – will allow us to meet these varied user needs.
We need your ongoing input to shape this exciting new service together!
- If you are a developer or adopter of AI, please complete this short form to register your interest in user research and receive further updates
- If you consider yourself part of the wider health innovation or AI support ecosystem and would like to be involved, please email the team at email@example.com.
- Future engagements might involve workshops / seminars, gathering content for the MAAS as well as joining up with complimentary initiatives
In the spirit of the MAAS collaboration, this blog was co-written by Clíodhna Ní Ghuidhir, (Principal Scientific Adviser, NICE) with inputs from various members of the MAAS working group.