The NHS AI Lab has provided a progress review for 2020 – 2021, including sharing key findings from its AI survey, as well as case studies and its plans for the future.
Focusing on AI technologies and AI-driven tools, the NHSX-hosted lab recently surveyed the UK’s AI community, before comparing the results to its 2019 edition – with the aim of finding which areas are seeing progress and which need support.
Described as a ‘sense-check’, it’s hoped the results will also help to shape future initiatives. Of 368 responses to the survey, 197 were AI developers for the healthcare sector.
One of the three main ‘key findings’ highlighted in its report was that diagnostics was once again the most ‘popular’ area of AI – with 57% of survey participants saying their technology focused on that area, compared to 34% for remote monitoring, 32% for triage and 25% for population health. However, the results are tempered by the fact that respondents could select more than one area in their answer.
Another notable finding from the NHSX report was that roughly half of the developers surveyed said they believed their AI products would be ready for deployment at scale in one-year – 24 points up from the previous survey. However, the majority of respondents were more confident their innovation would be ready in three (79%) or five (87%) years.
The NHS’s third key takeaway from the survey was that developers felt the COVID-19 pandemic had impacted progress in both positive and negative ways – with a third indicating negative repercussions, such as staff re-deployment and reduced data collection, while a similar number cited positives such as rapid uptake.
Other survey findings included only 18% of AI developers saying they had secured CE mark classification for their technologies – although 45% said they were in the process of doing so, while clinicians and patients with long-term conditions were found to be the main users of AI products. But what about ‘settings’ of use? Almost three quarters of AI technologies were discovered to be intended for use in secondary care. However, as technologies tend to be flexible and designed for use in multiple areas, 71% developers reported two or more point of care sites.
Building on those findings of AI-driven progress in diagnosing disease – the report highlights image analysis in areas such as CT scans and X-rays in particular as a ‘dominant’ area for AI, with further cancer and stroke screening techniques being funded by the NHS AI Lab’s AI in Health and Care Award.
HTN has reported on the development of several AI diagnostic tools over the past year, including a recent technique that uses AI algorithms to help improve treatment decisions for bowel cancer patients and a new AI diagnostics tool for Barrett’s Oesophagus, as well as achievements from Glasgow’s AI research centre.
In the wider AI field, specific case studies used in the yearly summary, also note the NHS AI Lab Skunkworks project ‘Data Lens’ – a fast-access data search that works in multiple languages, and the new SMART box which provides regional chest imaging data to boost the National COVID-19 Chest Imaging Database.
Other interesting studies flagged up took us back to diagnostics with breast cancer screening AI, which uses deep learning on millions of images to help clinicians analyse mammograms to increase speed and accuracy of diagnosis, as well as AI-driven smart phone urine testing for patients with diabetes and high blood pressure, to help detect potential early kidney disease.
“Through the AI Lab, we are seeing examples of how AI can play its part in helping to alleviate some of the pressures facing the NHS, particularly in light of the COVID-19 pandemic, whether it be through efforts to speed up diagnosis and treatment, or alleviating clinician time and assisting with early detection.
Concluding its summary of the year, the report also recounted its deliveries over the course of the last 12 months – which stretch from the funding rounds in the AI in Health and Care Award to skunkworks projects, and work in regulation, ethics and leadership.
Looking towards the future, the review states that UK has the ‘second highest number of AI-driven healthcare technologies in development globally after the US’, according to the NIHR Innovation Observatory.
It concludes that ‘access to data, the quality of it and the integration of data systems across the NHS needs to be improved if we are to further our international role’ and points to the future publication of a Data Strategy for Health and Adult social care by NHSX later in the year, which it says will set out ‘the evolution’ to a ‘truly data-driven health and adult social care system’.
Read the full 2020 – 2021 NHS AI Lab review online. And, if you enjoy reading about the future of artificial intelligence in healthcare, take a look at HTN’s ‘diverse applications of AI‘ series, too.
