The RCP report, based on survey responses and analysis of clinical experience, found that 70% of respondents see the inability to integrate AI tools with existing EPR systems as the main obstacle to adoption. Doctors stressed that the NHS’s current digital architecture is highly fragmented, with different EPR systems operating in isolation, making data sharing difficult and hindering the implementation of algorithms designed to support clinical decision-making. The Royal College of Physicians recommends that the government invest in a modern, well-functioning digital infrastructure, upgrade IT systems, and establish a central NHS-approved repository of algorithms, AI tools, and patient-facing applications that meet national standards.
More than two thirds of those surveyed (68%) said the NHS’s digital infrastructure is not ready for large-scale AI deployment, while 66% reported having received no training in AI tools, despite 79% saying such training is needed. These findings point not only to technical shortcomings but also to gaps in skills and preparedness for digital transformation across the healthcare system.
Dr Anne Kinderlerer even stated that “Physicians think the NHS is fundamentally unprepared for AI because its digital foundations are broken.”
The report highlights the lack of unified EPR standards within the NHS, meaning that laboratory results, clinical notes and demographic data are stored and presented differently depending on the organisation. This increases administrative burden for clinicians, slows down care and raises the risk of errors. The RCP recommends establishing a common EPR content model and clear interoperability standards, which it says are essential for enabling effective integration of AI into clinical systems.
Digital health experts note that EPR integration problems are not unique to AI but reflect broader challenges in healthcare digitisation. Fragmented systems, poor interoperability and inconsistent IT implementations have long been cited as barriers to fully realising the potential of electronic records, both for AI applications and for routine patient care.
The report concludes that without coordinated action on infrastructure, standards and clinical training, the deployment of AI tools in the NHS is likely to remain limited, despite growing interest in data-driven technologies across healthcare.