It is easy to run a pilot, yet hard to scale up, but a recent episode of It's Personal – the podcast series from Civil Service World and KPMG which is all about citizen experience – shows a practical way to make the transition possible. In conversation with CSW's editor-in-chief Jess Bowie, Rob Dale, head of AI diagnostics and implementation lead for digital histopathology at NHS England, explained how the £21m AI Diagnostic Fund was designed from the outset to avoid small trials and instead embed AI into everyday clinical practice with remarkable results: 3.2 million chest x-rays have been reviewed with the support of AI, abnormal case turnaround times have fallen from around eight days to four, and the programme is having an impact in every region of the UK, rather than only in the best-resourced trusts.
The conversation echoes what KPMG's Citizen Experience Excellence research has already demonstrated, namely that citizens want to be remembered, recognised and respected. Below are the key lessons from this episode, but it is worth listening in full and hear the story of a patient in the North East whose chest x-ray was flagged on the very first day the AI tool went live. [Podfollow link to episode]
Built for every region, not just the wealthy trusts
From the get-go, national reach was a non-negotiable objective of the fund. Dale explained that every region of the country had to benefit, not just the trusts that could afford to innovate, and that this commitment was a driving force behind the fund’s delivery. Rather than simply handing out money and checking back a year later, NHS England used existing imaging networks – groups of trusts that already worked together geographically – as the building blocks for deployment. This approached was paired with a central steer and a senior responsible owner at national level to keep momentum going across the whole programme.
That objective aligns with what citizens tell us they expect: to be remembered. As KPMG’s Jo Thomson noted, the research points to resolution, time and effort as pillars of citizen experience, and nothing brings that to life more clearly than knowing your case has been seen, prioritised and acted upon. Thomson, a former cancer patient herself, knows first-hand what it means to wait in a system that offers little information and no reassurance, with anxiety building by the minute. The AI Diagnostic Fund, she reflected, has been designed to help prevent exactly that experience.
Shared learning and the human in the loop
If shared purpose set the direction, what got the programme moving at pace was challenging the siloed culture that has shaped relationships between NHS trusts for years. Dale described how trusts and networks were brought together as a community, sharing documentation, approaches and lessons learned as standard practice. Trusts that were further ahead were expected to share their materials so that others did not have to start from scratch, which is a significant culture shift. This approach allowed trusts to learn from one another rather than working things out independently, and deployment moved faster, so that more patients across the country could benefit sooner rather than waiting for their local trust to catch up.
Equally important was the recognition that AI's first job is to support the people doing the work, not replace them. Dale explained that every image flagged by the AI tool is still reviewed and reported on by a trained clinician, and that the technology's role is to help prioritise and triage, so that the clinical judgement and the human relationship with the patient remain in human hands.
Transparency at every level
Dale addressed the question of how much patients should be told about the role AI plays in their care. He explained that NHS England deliberately left it for each trust to decide, and that organisations have responded differently. For example, some trusts have run local media campaigns and put up posters as part of their communication strategy, and others mention it to the patient at the point the x-ray is taken. He said that patients should never feel that something is being hidden from them, and the information about how AI is being used should always be findable, whether through a leaflet, a website or a conversation with staff.
However, Dale also warned of AI becoming “its own cottage industry”, with organisations feeling obliged to tell people about “everything that we’re doing” with the technology. Some patients, he noted, responded to being told AI had been used with a shrug, saying: “I thought you were doing that anyway.” Dale suggested such reaction is a sign of how quickly AI is becoming part of everyday expectation.
That approach reflects respect for the citizen: treating patients as people who deserve transparency and the ability to ask questions.
Made to measure
Dale offered several hard-won lessons for other parts of government, but perhaps the most important is also the most straightforward: decide before you start how you will measure success and make sure the means of measuring it are in place from the beginning. NHS England can now point to abnormal case turnaround times falling from around eight days to four, and overall turnaround times falling from five days to four, precisely because those metrics were defined early and tracked consistently across the country.
The other lesson is about confidence. There will always be naysayers who insist that a scaled deployment of new technology across a complex organisation like the NHS simply cannot be done. The AI Diagnostic Fund (which has reached 64 trusts, exceeding its original target of 40 to 50, and now supports a third of all chest x-rays across the NHS) is proof that it can.
Listen to the full episode of It’s Personal, available now wherever you get your podcasts.