The UK’s Labour government has ambitious plans to transform the country’s National Health Service (NHS) into one that is more of a preventative than a sick care service. A large part of meeting this aim is centred around making the NHS the “most artificial intelligence (AI)-enabled” health system globally.
To meet this aim, the government’s 10-year plan for the NHS has pledged to bolster areas such as diagnostics, screening, and patient monitoring by deploying AI to improve upon existing processes.
Go deeper with GlobalData
Discover B2B Marketing That Performs
Combine business intelligence and editorial excellence to reach engaged professionals across 36 leading media platforms.
Meanwhile, the UK Medicines and Healthcare products Regulatory Agency (MHRA) is reevaluating its stance towards how AI is regulated. The agency’s National Commission recently published its final recommendations from stakeholders that will contribute towards the development of a new regulatory regime for AI products and medical devices that are intrinsically linked to the UK government’s 10-year plan for the NHS.
Against the backdrop of these ongoing shifts, at the 2026 NHS Confed Expo, which took place on 11-12 June in Manchester, UK, AI and the practical challenges around its effective implementation across the NHS were in the spotlight.
Creating the conditions for successful AI screening
During a panel session on NHS Confed’s collaboration and partnerships stage, the challenges and opportunities for AI as a screening tool were under scrutiny. Judicious implementation of valuable use cases and collaboration efforts surrounding the technology emerged as key considerations.
Reflecting on the importance of collaboration, Professor Ben Bridgewater, executive chair of the Manchester-based Health Innovation Network, said: “What we’re short of is a good method for taking impactful technologies like AI and scaling them up to have the maximum impact for patients, NHS productivity, and the wider economy, and no organisation can do that alone.”
Sarah Woolnough, CEO at The King’s Fund, noted that adoption of AI across the NHS and the UK’s broader healthcare sector will be reliant on effective leadership.
The UK charity recently interviewed over 60 healthcare leaders on the challenges of AI adoption and implementation, and found that creating the conditions for success was a recurrent sentiment.
“Too often we might fund the technology, but we don’t fund the change,” Woolnough explained.
“Because the pace of change and innovation is so rapid, we must pay attention to the conditions for success. By this, I mean that we have to think about the human as well as the technical aspect and ensure that people leaders are as involved in driving change as tech leaders,” Woolnough continued.
Regarding AI’s application as a screening tool to support earlier diagnosis, panellists agreed that work being undertaken in this area must not happen in a vacuum if AI is to become part of standard practice across the NHS for a range of diagnostic modalities.
Speaking on the same panel, Colette Marshall, CEO of Diabetes UK, addressed AI’s role in the earlier diagnosis of type 1 diabetes (T1D) in children. In primary care, Marshall highlighted that many children with the condition are diagnosed late, with around 25% diagnosed in a serious clinical condition, as per clinical records, or 45% on the basis of patient reported experience measures (PREMS) data.
Being diagnosed late or not diagnosed at all increases the risk of mortality in this patient population, but “one surgery may never see such a patient, so how do you try and raise awareness for something that people may not see?” Marshall asked.
To investigate this problem, Diabetes UK funded a 2024 study in Cardiff, Wales in which an AI tool was trained on data from one million children to detect patterns in their GP records that could signal they have undiagnosed T1D.
The research found that the AI tool could successfully alert doctors in advance for 72% of children who would go on to develop T1D in the following 90 days.
According to the research, which was published in the Lancet Digital Health, on average, the AI tool would have allowed children to be diagnosed and started on insulin therapy nine days earlier than they were – a timescale that could in some cases represent the difference between life and death.
“I think what we’re seeing in T1-T2 diabetes and in the diagnosis of complications means that different parts of the system are going to have to work together in very different ways, because data may be collected in one part of the system that’s liable to inform a diagnosis elsewhere,” Marshall explained.
“Regarding collaboration in AI, this is going to be an interesting question for us,” she concluded.
Returning to AI governance, Bridgewater emphasised the importance of owning AI at the “top of the shop”, with a rigid operational model in the order of people, process, culture, tools, and tech as means for effective transformation.
“We’re absolutely rigorous around that [the order of the operational model], because for AI to deliver screening at scale, you need to change all of those things,” Bridgewater explained.
“You need to have people with different capabilities, and clear processes, which are properly mapped out, and suitable for transformation with AI of whatever sort you’re talking about. But the culture needs to be fit for purpose and embrace the technology accordingly. Meanwhile, it’s important to have the right tools and tech to support this overall shift.”
Adding to the governance aspect, Woolnough highlighted that the ‘AI-powered patient’ is something we need to “get a grasp of”, with appropriate patient pathways designed accordingly.
“That will change the diagnosis of disease, it will change our ability to prevent, and across healthcare in the UK currently, we’re likely not thinking about AI in as transformational a way as we could be,” she concluded.
AI models: the importance of building trust
During a panel session on NHS Confed’s AI theatre stage, speakers focused on the growing visibility of AI in healthcare.
With the rise of AI chatbots like ChatGPT, many people now use AI on a daily basis. Regarding healthcare, AI engagement is beginning to impact how people interact with the healthcare system. Research published by Gallup in April 2026 revealed that 14% of recent US chatbot users reported that AI-generated information or advice they received led them to skip a healthcare provider (HCP) visit in the past 30 days.
Projected out to the entire adult population, this means that around 14 million US adults did not see a HCP due to AI-generated health information or advice they received instead.
However, for AI models used in healthcare, trust is of paramount importance, with the panellists agreeing that trust in freely available chatbots’ dispensation of advice should be taken with caution.
According to Dr Mohammad Al Ubaydli, CEO of digital patient health record company Patients Know Best, AI, especially online chatbots, should be viewed with scepticism because they often receive incorrect or incomplete data that affects the conclusions they make.
“As long as we’re feeding the correct datasets or a single patient record from all sources of data about a patient or doctor asking questions [to an AI], we then you have a good foundation for the AI to improve its safety and efficacy,” Al Ubaydli noted.
In addition, Susan Thomas, clinical director at Google Health, explained how Google approaches the challenge of ensuring its AI tools are safe and useful.
“In our health team, we have stakeholders including clinicians, doctors and nurses, and a big part of our job is focused on evaluating how the models work,” she said.
“This involves building clinical evaluation sets, robust datasets to test AI models, and then evaluating them to ensure that the answers are helpful, safe, and aren’t harmful in any way.”
Dr Richard Whittington, partnerships director at Ufonia, a company that builds AI assistants, asserted that trust in AI systems is not about AI itself, but the context in which the technology is used.
With the current healthcare model of the NHS, Whittington noted that patients typically expect to see a general practitioner (GP).
He said: “Seeing a GP is entirely appropriate in some cases, but in others, it may be more appropriate to see a nurse or healthcare assistant. I think that’s where we’ll go with AI models.
“I think it will start by moving patients across, establishing AI as a viable pathway for healthcare delivery in the future. And, eventually, it will become natural.”
The UK government is focused on AI as a key tool to make the NHS better. But it remains to be seen how its various intentions will be implemented. What is clear, however, is that to achieve digital transformation across the NHS, with AI as a central tenet, close collaboration will be critical. Meanwhile, care must be taken to ensure that AI models deployed across the health service not only provide meaningful benefits but do so in a safe manner.
