Data collection, compatibility with legacy systems, governance and contracts are all pose automation problems for the UK’s public health service.
Published in The Lancet eClinicalMedicine, a new study by University College London [UCL], the Nuffield Trust, and the University of Cambridge, has exposed significant barriers to implementing artificial intelligence across the NHS.
Those responsible for the research hope it will ‘provide timely and useful learning’ for the government. Downing Street recently unveiled a 10-year plan for digital transformation within the health service, with big plans to improve service implementation and treatment delivery through AI.
An early example of this is a programme which looks to automation for the diagnosis of chest conditions, including lung cancer, across 66 NHS hospital trusts in England. Coating £21million in public funding, the goal is to improve access to specialist opinions, while critical cases will be prioritised for specialist review and abnormalities will be highlighted on scans to support specialist decisions.
Prior studies, which were predominantly lab-based, had suggested AI could benefit diagnostic decisions, including detection accuracy, and ease the staff burden. However, the UCL-led study has found AI tools took longer to implement than anticipated by leadership, with timeframes for contracting between four and 10 months longer than predicted.
‘Our study provides important lessons that should help strengthen future approaches to implementing AI in the NHS. We found it took longer to introduce the new AI tools in this programme than those leading the programme had expected,’ said study first author Dr Angus Ramsay of UCL’s Department of Behavioural Science and Health.
As of June 2025, one third of the hospital trusts involved were still not using the tools in clinical settings, despite the programme having run for 18 months. A number of specific issues were identified, including limited opportunities to engage clinical staff due to their existing workloads, embedding AI technology within existing NHS IT systems, and a general lack of understanding and scepticism surrounding artificial intelligence.
In many cases, those responsible for procurement reported being overwhelmed by technical information, raising the risk of important details being missed — a factor that needs to be overcome. It has also been recommended that the government should create a list of potential suppliers to expedite procurement at a local trust-by-trust level.
Staff being asked to use the AI tools were also reluctant in many cases, with significant concerns around where accountability would fall in the event of missed or misdiagnosis. Training failed to address this properly.
In contrast, advice and training was available from a central team at the procurement stage. Researchers also observed high levels of commitment and collaboration between local hospital teams — including clinicians and IT consultants — and AI suppliers to improve implementation.
‘In this project, each hospital selected AI tools for different reasons, such as focusing on X-ray or CT scanning, and purposes, such as to prioritise urgent cases for review or to identify potential symptoms,’ said senior author Professor Naomi Fulop of UCL’s Department of Behavioural Science and Health.
‘The NHS is made up of hundreds of organisations with different clinical requirements and different IT systems and introducing any diagnostic tools that suit multiple hospitals is highly complex,’ she continued. ‘These findings indicate AI might not be the silver bullet some have hoped for but the lessons from this study will help the NHS implement AI tools more effectively.’
Image: Jair Lázaro / Unsplash
More Digital Business:
4Chan and Kiwi Farms sue UK Government over Online Safety Act
Leave a Reply