NHS AI Readiness Landscape 2026

Abstract

# Project Title: NHS AI Readiness - Scoping the State of the Nation ## Background and rationale Interest in artificial intelligence in healthcare is growing quickly, but we know little about how prepared United Kingdom (UK) National Health Service (NHS) organisations are to adopt it safely and at scale. National strategies and pilot projects tend to feature a handful of well-resourced centres. That leaves open how the rest of the NHS is placed in terms of leadership, planning, governance and workforce. Without a systematic picture, it is hard to target investment, spread good practice, or judge whether the NHS can support collaborative approaches such as a federated AI network, where models are trained across sites without patient data leaving the organisation that holds it. This project takes a national snapshot of AI readiness across NHS trusts in the UK. ## Purpose and research question We define readiness as **a combination of capability and motivation**: whether a trust has the structures, people and policies to use clinical AI responsibly, and whether it has the intent to do so. The primary question is how ready NHS trusts are to adopt and govern clinical AI. Our working hypothesis is that there are huge variations in clinical AI readiness in the UK within its health system. A secondary goal is practical. We want to identify trusts with the potential to join major AI networks, so that the findings feed directly into collaborative work as well as describing the current position. ## Methods The study uses Freedom of Information (FOI) requests sent to NHS trusts. This gives a consistent, auditable route to organisational-level information that is rarely published. The questions are six simplified items adapted from the DIME Health AI Readiness Assessment, each covering one domain of readiness: 1. **AI leadership:** whether there is a named executive or clinical lead for AI. 2. **AI strategy and planning:** whether AI features in the trust's strategic and operational plans. 3. **AI policy:** whether formal policies exist for AI adoption and use. 4. **Governance and oversight:** whether there are structures for approving, monitoring and assuring AI tools. 5. **Use of AI in clinical care:** whether AI is currently deployed in clinical settings. 6. **AI literacy and training:** whether staff are offered training and development. ## Planned analysis After collation, we will analyse the data to look for patterns in readiness by: - **Region**, to identify geographic variation. - **Trust type**, comparing teaching hospitals with district general hospitals, acute with non-acute trusts, and Foundation with non-Foundation trusts. - **Trust size**, using staff numbers, catchment population and, where relevant, bed numbers. Analysis will begin with descriptive summaries of each readiness domain, followed by comparisons across the trust characteristics above. Trusts that score consistently well across domains will be flagged as potential exemplars for the secondary goal. We will also consider whether other factors need to be examined. Detailed prospective analysis plan will be added here with this registration before completing all data collection. ## Expected outcomes - **A national baseline** of AI readiness across NHS trusts, showing which domains are most and least developed. - **A test of the hypothesis** that dedicated personnel, strategy, policy, governance and training are largely absent, with evidence of where the gaps are widest. - **Insight into what explains variation**, for example whether readiness clusters by region, trust type or size. - **A shortlist of exemplar trusts** that could anchor or join a federated AI network. - **A peer-reviewed manuscript** setting out the findings. Journals under consideration include JMIR and its sister titles (JMIR Human Factors, JMIR AI), PLOS Digital Health, the International Journal of Medical Informatics, BMJ Health & Care Informatics and BMC Health Services Research. - **A foundation for future work**, including follow-up studies and network-building activity. ## Significance The findings should help NHS leaders, commissioners and policymakers see where the practical barriers to AI adoption lie, and give them an evidence base for targeting support and investment. By identifying trusts that are already well prepared, the project also gives collaborative approaches such as federated learning a realistic place to start.