The National Institute for Health and Care Excellence (NICE) has published a new delivery plan setting out how it intends to support the safe, effective and scalable adoption of artificial intelligence across the NHS.
The plan focuses on how AI can be used to generate and evaluate healthcare evidence, how AI-powered technologies should be assessed for NHS use, and how NICE can responsibly deploy the technology within its own operations.
Writing alongside the plan, NICE Chief Executive, Professor Jonathan Benger said that artificial intelligence was rapidly moving “from promise to practice”, with tools already being used to support diagnosis, interpret medical images, identify patients at risk of illness and reduce administrative pressures.
However, he emphasised that the central challenge was not whether AI would become part of healthcare, but how it could be introduced in a way that improved care while remaining safe, effective and fair.
Building public confidence through evidence
NICE said its approach would support the Government’s ambition to harness the opportunities created by artificial intelligence, while recognising that public confidence must be earned through transparency, evidence and robust standards.
The organisation’s work during the next year will be divided into three principal areas:
- using AI in evidence generation and the evaluation of innovation;
- evaluating health technologies that incorporate AI; and
- exploring how AI can improve NICE’s own productivity.
A central element of the plan is the principle that healthcare decisions must continue to be guided by credible evidence, regardless of whether artificial intelligence is involved in producing or interpreting that evidence.
AI has the potential to analyse large quantities of information more quickly than traditional research methods. It could help researchers identify patterns in data, review scientific literature, synthesise findings and analyse real-world healthcare information.
NICE warned, however, that speed must not be prioritised at the expense of reliability.
Evidence used to support decisions about patient care must be transparent, reproducible and explainable. Researchers, clinicians and decision-makers must be able to understand how conclusions have been reached and whether the underlying methods are sufficiently robust.
New AI best practice framework due by March 2027
By March 2027, NICE intends to publish an AI best practice methods framework covering the use of artificial intelligence in evidence generation, synthesis and review.
The framework will build upon testing undertaken through the NICE Health Technology Assessment Innovation Lab, as well as internal work examining how AI could support evidence evaluation and the development of guidance.
NICE will also seek to improve AI knowledge and capability among its staff and external partners involved in producing guidelines and assessments.
The work could establish clearer expectations for developers seeking to use AI-generated or AI-assisted evidence when demonstrating the clinical and economic value of new products.
AI technologies must prove that they improve care
The second strand of the plan focuses on the evaluation of health technologies that incorporate artificial intelligence.
These could include clinical decision-support tools, diagnostic software, predictive systems that identify patients at greater risk of disease and digital technologies that help people manage long-term conditions.
NICE highlighted diagnostic imaging systems and tools that detect early signs of patient deterioration as examples of technologies that could support earlier intervention and improve the use of clinical expertise.
However, the organisation stressed that innovation should not automatically be treated as evidence of effectiveness.
AI-enabled products will need to demonstrate that they improve patient outcomes, provide good value for public money and perform consistently across different communities, care settings and population groups.
Patients and healthcare professionals must also have confidence that AI tools are safe, trustworthy and appropriate for use in real clinical environments.
NICE already offers rapid and lighter-touch evaluation routes for some early-stage digital and medical technologies that address national priorities. This has included recent guidance covering five artificial intelligence tools designed to support the detection of fractures.
NICE and MHRA to coordinate on regulation and evidence
NICE said it would work more closely with the Medicines and Healthcare products Regulatory Agency (MHRA) to coordinate regulatory requirements and expectations around evidence.
This could help developers understand more clearly what information will be required to satisfy both regulatory assessments and evaluations of clinical and cost effectiveness.
NICE will also undertake technology appraisals of AI-enabled HealthTech and publish four guidance documents covering AI technologies referred to the organisation by the Secretary of State.
Closer coordination between NICE and the MHRA is likely to be particularly important for technologies that change over time, depend upon continually updated data or behave differently when deployed across different patient populations and NHS settings.
AI to support faster NICE guidance
The third priority is to explore how artificial intelligence can improve the way NICE itself operates.
Potential uses include summarising large volumes of information, identifying relevant evidence more quickly and streamlining administrative processes that currently require substantial staff time.
NICE has set a target of halving the staff time required to produce each piece of guidance by 2030.
Its HealthTech guidance-producing directorate will be the first test case for embedding AI within a new delivery model.
The organisation said any use of artificial intelligence would need to be carefully governed, ensuring that gains in productivity did not weaken the quality, independence or transparency of its recommendations.
NICE is also developing a wider AI Framework that will clarify how it develops, evaluates and uses artificial intelligence to generate evidence.
Welcoming the announcement, Chair of the life sciences and healthcare working group at UKAI, Adama Ibrahim said: “This is a highly encouraging and important strong signal from NICE. Sending a strong signal that the UK is committed to enabling responsible AI innovation while maintaining rigorous standards of evidence, safety and value. Innovation alone is not enough. AI must demonstrate measurable clinical benefit, value for taxpayers and consistent performance across diverse patient populations.
“Close alignment between the MHRA’s regulatory approval process (focused on safety, quality and performance) and NICE’s health technology assessment (focused on clinical effectiveness and cost-effectiveness) will be critical to creating a predictable evidence pathway that enables safe innovation to reach patients faster while giving developers, clinicians and the NHS greater confidence in adoption.
“As collaboration evolves, there is also an opportunity to further integrate these evidence pathways. For adaptive AI-enabled medical devices in particular, where algorithms continue to evolve, a joined-up framework across regulation and reimbursement could become a significant competitive advantage for the UK. This is exactly the type of progress needed to strengthen the UK’s position as a global leader in trusted AI for healthcare, and I encourage everyone with an interest in AI-enabled health technologies to read NICE’s vision and the direction of travel it sets out.”
Linking evidence with NHS procurement
NICE’s delivery plan also complements the new NHS Shared Business Services Healthcare AI Solutions Framework, which was explored during a recent webinar hosted by UK Healthcare and Life Sciences Innovation (UKHLSI), NHS SBS and UKAI.
The framework is designed to provide NHS organisations with a compliant national route for procuring artificial intelligence technologies across areas including diagnostic imaging, pathology, predictive analytics, operational efficiency and integrated AI solutions. During the webinar, NHS SBS outlined how the framework could reduce fragmented purchasing arrangements and lengthy procurement processes while maintaining appropriate requirements around regulation, accreditation and patient safety.
The two initiatives address separate but closely connected barriers to NHS adoption. The NHS SBS framework can make it easier for trusts and integrated care boards to identify and procure suitable technologies. NICE’s work will help establish whether those technologies are supported by credible evidence, improve patient outcomes, represent value for money and perform fairly across different populations and healthcare settings.
Together with MHRA regulation, these developments could create a clearer pathway from innovation to implementation: regulatory assurance from the MHRA, evidence-based evaluation from NICE and a compliant procurement route through NHS SBS. However, securing a place on a procurement framework should not be viewed as the end of the assessment process. AI tools will require ongoing monitoring after deployment to determine whether they continue to perform safely, release clinical capacity and deliver measurable benefits in real NHS pathways.
Innovation must demonstrably improve care
Dr Mark Ratnarajah, Clinical Director of Strategic Accounts at Sword Intelligence and Chair of the UKHLSI AI Working Group, welcomed NICE’s focus on evidence, transparency and measurable improvements in patient care.
He said:
“It is encouraging to see NICE set out such a clear, evidence-led plan for AI in the NHS.
“The focus on transparency, reproducibility and proven outcome improvement – not just speed – mirrors exactly the bar we hold ourselves to at Sword as we deploy AI-driven care management across NHS trusts and ICBs in the UK and EU.
“Innovation only matters if it demonstrably improves care and works fairly across populations. We welcome the coming AI best practice framework and closer MHRA coordination and look forward to contributing our deployment evidence to help shape it.”
From experimentation to responsible adoption
The delivery plan reflects a wider shift in the debate around healthcare AI, from whether the technology can be used to how effective tools can be adopted safely and consistently across the health service.
While AI could improve productivity, support clinicians and help patients access earlier interventions, implementation will require credible evidence, appropriate regulation and ongoing monitoring of performance after deployment.
NICE said it would continue working with patients, clinicians, innovators and partners across the health and care system as its approach develops.
Professor Benger said the organisation’s objective was to combine innovation with evidence, ambition with accountability, and technological progress with public trust.
The challenge will now be to translate those principles into evaluation processes that can keep pace with rapid technological change while ensuring that patient safety, fairness and clinical value remain at the centre of NHS adoption.
Find out more
To find out more about UK Healthcare and Life Sciences Innovation (UKHLSI) AI Working Group and UKAI’s Life Sciences and Healthcare working group, please contact partnerships director, Ben McDermott at ben.mcdermott@chamberuk.com.