NICE is working in partnership with the Care Quality Commission, Health Research Authority and Medicines and Healthcare products Regulatory Agency to deliver a multi-agency advisory service (MAAS) for developers and adopters of data-driven technologies, such as artificial intelligence (AI).
This innovative new cross-regulatory service will provide support on the regulation and evaluation of AI technologies in health and care. By enabling the development and widespread adoption of safe, innovative, value-adding technologies, the service will play a key role in optimising care and improving patient outcomes.
There will be two elements to the service. Firstly, an informational website will provide centralised, curated and up-to-date regulatory content for both developers and adopters of AI and data-driven technologies. Secondly, a transactional side, where developers and adopters will be able to access specialist support from the MAAS partners.
I’ve been project manager for the MAAS since November 2020 and it’s a fantastic project to be involved in. Although working collaboratively across four different organisations does have its challenges, we have implemented mutually agreed ways of working and all work together well. It’s rewarding to see everyone moving forward and taking progressive steps together.
User research
As part of our development of the MAAS web platform, we must adhere to the Government Digital Services assessment pathway. This helps teams to create and run great public services that meet user needs. We recently passed the first phase of this - the alpha phase. The assessment panel was particularly impressed with our work to understand users and their needs.
Undoubtedly, user research is one of the most important aspects of the whole project. The MAAS partners are experienced and knowledgeable, but they are not developers or adopters. To create a truly user-centric service, we have actively engaged with these groups from the very beginning.
As well as organising the practicalities of the user research, I was responsible for recruiting a broad range of participants, corresponding with them, posing questions, taking notes and observing the various sessions. It was fascinating to be a fly on the wall and we’ve made a number of significant changes as a result of their feedback.
For example, we originally looked at categorising particular pieces of regulation or guidance as ‘must’ or ‘should’ do. But users told us this was confusing. They were also likely to ignore the ‘should’ advice because to them, the wording suggested they did not have to follow it. We took this feedback on board and have now changed the wording to ‘legal requirement’ and ‘best practice’.
Here’s a helpful summary of our top-level user-research findings:
- There's a diverse range of potential users and most of them struggle to navigate the regulatory landscape in health and care.
- Large and established developers have dedicated regulatory teams whereas small and new developers tend to turn to regulatory consultants for support.
- Users would benefit from a central ‘front door’ that summarises all key regulatory and best practice guidance in a standardised way and signposts to trusted sources.
- Arranging guidance according to high-level development and adoption pathways supports users in understanding what they need to consider now and in the future.
- Ensuring that the language used is simple, clear and consistent is essential to making the service useful and usable.
Get involved
There’s still lots of work to do if we are to hit our target launch date in early 2023 – and we need your help.
Over the coming months, we will begin building a web platform for the service. We need potential users to work with us to test, develop and improve it. So, we're inviting developers and adopters of data-driven technologies, such as AI to register to help shape the new service. We want to make sure we capture a broad range of perspectives from all our potential audience groups in our user research, so that the service is inclusive and useful for everyone.
If you have any questions, please get in touch by emailing ai.advice@nice.org.uk.
Blog comments
There are no comments