Transformation Directorate

2020-21: A year in the life of the NHS AI Lab

Read a review of the progress of the NHS AI Lab during 2020. Find out what the Lab has achieved and the work in progress for AI in health and social care.

Artificial Intelligence (AI) technologies can make a significant difference to the health and social care system, and most importantly to patients. AI-driven tools can be used to analyse large quantities of complex information to detect patterns and predict outcomes. These tools have the potential to change and improve the way that health and adult social care services are provided to people by improving the quality and efficiency of care.

The NHS AI Lab was set up to make the most of the potential of AI technologies to transform health and adult social care. Despite COVID-19, this past year has brought about great opportunities as well as challenges for the use of AI in health and care.

To further explore what the impact of those opportunities and challenges have been on the development of AI in healthcare, we recently surveyed the UK’s AI community. Comparing the results of this survey with those from 2019 informs us about the technologies that are being developed, the progress made, and the support needed to accelerate this. The results are a useful sense-check for the programmes of the NHS AI Lab and help with planning our future objectives.

Regularly reviewing the state of AI technology development and use helps us to be confident that our work is achieving what is needed: to bring real health and care benefits to the people who need it most.

These are 3 of our key findings:

1. Diagnostics is the most popular area for AI in health and care

Areas for AI

The deployment of AI-driven technologies is not yet evenly spread across the different areas of health and social care. The figure shows the responses from survey participants who were asked about the area of focus of their technology. Survey participants were allowed to select more than one area of focus. The survey clearly shows that most of the early adopters of AI technologies are in diagnostics. Similarly to 2019, the 2021 survey indicates the use of AI is dominant in the following 4 areas:

  • Diagnostics - 57%
  • Remote monitoring - 34%
  • Triage - 32%
  • Population health - 25%

2. AI products have some way to go before large scale use

Ready for deployment

Although the levels of readiness for the deployment of AI technologies in health and care have significantly increased from 2019 to 2021, it is still in early stages.

Around half of AI developers surveyed in the UK believe their product will be ready for deployment at scale in one year; up 24 percentage points from the last survey.

  • Ready in 1 year: 54% developers
  • Ready in 3 years: 79% developers
  • Ready in 5 years: 87% developers

3. The COVID-19 pandemic has influenced progress

Pandemic influence

The pandemic has had an almost equal positive and negative effect on the development and adoption of AI technologies into health and care settings.

A third of AI developers in the survey indicated a negative impact, giving examples of problems with re-deployment of clinical staff, reduced data collection and lack of engagement for non-COVID-19 activity.

However, a similar number indicated positive impacts where healthcare pressures had resulted in a rapid uptake of AI tools and an increased acceptability for digital technologies being used to deliver care.

More findings from the 2021 AI survey

The survey results represent a snapshot of AI developers and procurers each year. The survey is open to all and received over 368 responses in 2021, 197 of whom were developers of AI for health and care.

18% of developers had secured CE mark classification for their technology - read more

Only 18% of AI developers indicated they had secured medical device CE mark classification for their AI-driven technology, remaining steady from 2019. This was also reflected in the National Institute for Health Research (NIHR) Innovation Observatory’s Mapping the Global Activity of AI Health Technologies (2021), who found that overall only 21% of AI-driven technologies identified had obtained regulatory approval.

In the 2021 survey, a further 45% of AI developers indicated they were in the process of securing classification, an observed increase of 15% from 2019. Knowledge of the move from CE to UKCA marking is low, with half of AI developers unaware of the change that came into effect on 1 January 2021. Combined, the above suggests a need to raise awareness on the current and upcoming regulatory changes.

Three quarters of AI technologies are intended for use in secondary care - read more

Nearly three-quarters of AI-driven technologies were aiming to be deployed in secondary care, followed by primary care and community care. The majority of AI-driven technologies are designed for use in more than one of these areas, meaning potential benefits for people accessing everything from GPs to specialist services. 71% of AI developers report two or more point of care sites.

Clinicians and people with long-term conditions are the main users of AI products - read more

Clinicians and patients with long-term conditions continue to be the most prevalent user groups for AI-driven technologies (e.g.clinical decision support tools or remote tracking and monitoring of symptoms).

According to the NIHR Innovation Observatory’s Mapping the Global Activity of AI Health Technologies (2021), the most common indications in AI development are oncology, cardiology, and neurology.

How is the NHS AI Lab helping AI-driven technology development and adoption?

As demonstrated in this year’s survey, diagnosis of disease - in particular the reading of images like CT scans and X-rays - is proving a dominant area for the use of AI-driven technologies . The NHS AI Lab’s AI in Health and Care Award is funding the testing and research for a range of AI products that aid diagnosis, like screening for cancer, assessing stroke images and much more.

Image analysis is not the only area to benefit from the advance of AI technology. New products are being trialled that will automate and streamline patient scheduling and triage. Testing is in progress for monitoring software that people can use at home, minimising unnecessary trips to GPs and hospitals.

The NHS AI Lab’s programmes support new technologies like these from start to finish. The NHS AI Lab specialises in:

  • Solutions - testing the feasibility of early stage ideas for applications of AI in healthcare through our Skunkworks
  • Funding - supporting research and trials through the AI Award
  • Datasets - providing medical imaging data for use by AI developers in our Imaging programme
  • Equality - guiding the development of AI that is fair for everyone through the AI Ethics initiative
  • Approval - giving developers a clear route to comply with UK regulatory standards via our Regulations programme

What we have delivered this year

The NHS AI Lab has progressed multiple programmes of work and created partnerships with leading organisations throughout the AI landscape in order to find ways to deliver benefits to hospitals, clinicians, researchers, developers and the public through the following:

AI in Health and Care Awards

Working with over 70 NHS sites through our AI Award to research and develop AI-driven technologies for health and care. The Award is supporting technologies from round 1 to test and evaluate their benefit to the health and care system before being rolled out into widespread use. Round 2 awards will be made in May 2021.

AI in imaging

Collecting more than 38,000 images from 20 NHS trusts for the National COVID-19 Chest Imaging Database. This has enabled 13 projects to develop new AI technologies that will help to speed up the identification, severity assessment and monitoring of COVID-19.

AI Skunkworks

Establishing 4 skunkworks projects trialling AI ideas that may benefit the NHS. Our AI Skunkworks supports taking ideas from scratch to initial prototype and focuses on problems that will make a real difference for healthcare professionals, public health and social care.

Regulation of AI

Funding projects with partner health and care regulatory organisations to help AI developers, buyers and users understand how the regulation and evaluation of AI-driven technology works.

AI Ethics

Ensuring that AI development is done in a way that is fair and inclusive for everyone who is affected by it. Through our AI Ethics Initiative we are working with partners, such as the Ada Lovelace Institute and the Health Foundation. This work is looking to understand how to prevent algorithmic bias in AI technologies and how developments in AI can be used to improve the health and care experiences of minority communities.

AI leadership

Publishing policies and papers that will help establish the UK as a global leader on best practice for developing and implementing AI technologies for health and care, including the internationally recognised white paper, AI for healthcare: creating an international approach together, on behalf of the Global Digital Health Partnership.

What’s next for the NHS AI Lab?

Chest scans
Chest scans continue to be collected for the COVID-19 database

Working with our partners and the wider NHS and care sector, the NHS AI Lab will continue to take a leading role in providing guidance and real-world examples on where AI-driven technologies could be most useful in health and care.

The UK has the second highest number of AI-driven healthcare technologies in development globally after the US (according to NIHR Innovation Observatory’s Mapping the Global Activity of AI Health Technologies, 2021). It is our aim to continue creating an environment that enables both developers and adopters of AI technologies to thrive, and to bring the benefits of AI quickly and safely to the people who need it most.

There is an immense pool of knowledge and experience at hand in the regulators and bodies with whom we are partnering to achieve this, and within industry. The continuation of this joined-up approach is a principal factor in our future plans.

Access to data, the quality of it and the integration of data systems across the NHS need to be improved if we are to further our international role. NHSX will be publishing a Data Strategy for Health and Adult social care later in the year that sets out the evolution to a truly data-driven health and adult social care system.

To give you a flavour of what the NHS AI Lab is doing in 2021, we intend to:

  • continue supporting the pandemic response through increasing use of the National COVID-19 Chest Imaging Database
  • collaborate with other NHS programmes, government and the health and care sector to ensure that progress in the use of AI is part of the wider innovation and transformation of the NHS
  • work with a steady supply of early stage AI technologies to get them tested and trialled within 1-2 years
  • help the most promising technologies to get approved, commissioned and into widespread use in the NHS
  • develop a National Medical Imaging Platform to provide developers and researchers with high quality data that reflects a wide population base on a large geographical scale
  • work with the health and care sector to investigate the potential of using AI to solve a variety of problems, by testing and prototyping rapid skunkworks projects
  • educate and share knowledge with others who are developing and using AI in health and care so that progress is faster, quality is improved and costs reduced
  • update the UK’s regulation system for AI tech so that developers understand how to get approved and the public can be confident of safety and quality
  • make sure that progress in AI goes hand in hand with a commitment to ensuring all new technologies are safe, fair and ethical
  • set the vision for what the future of AI in health and adult social care could be

Further reading