Background
NHS hospitals face major challenges with long waits in A&E and for planned treatments. In March 2025, only around 6 in 10 patients in major A&E departments were seen within the four-hour target, and nearly 47,000 people waited more than 12 hours after doctors decided they needed admission. Broader measures show more than 1.6 million people experienced 12-hour waits from the moment they arrived at A&E in 2022. These delays are not just uncomfortable — they are linked to worse outcomes, with studies showing that long waits can directly increase the risk of avoidable deaths. Planned care is also under strain, with over 7 million people waiting for hospital treatment, including cancer and heart patients where delays of even a few weeks can affect survival.
Novelty & Importance
This project will explore the idea of an “AI Hospital of the Future.” Instead of focusing on a single task, it will bring together many different artificial intelligence systems trained for specific jobs — such as GP consultations, reading X-rays and scans, or interpreting heart and brain signals — and link them into a full virtual hospital. Using two of the UK’s richest health data resources, the Clinical Practice Research Datalink (covering tens of millions of GP records) and the UK Biobank (a half-million volunteer study with detailed health, lifestyle and imaging data), we will test how such an AI hospital might have managed real patients in the past, and whether it could have reduced delays or improved outcomes.
Aims & Objectives
– Develop specialist agents for GP triage/referral, radiology (XR/CT/MRI/PET), cardiology (ECG), neurology (EEG), and MDT/tertiary decisions.
– Build retrospective simulators to replay historical cases and compare AI vs. actual outcomes.
– Quantify diagnostic accuracy, time-to-diagnosis, referral appropriateness, length-of-stay and fairness.
– Produce NICE ESF and MHRA AIaMD-aligned evidence to inform any later NHS pilots.
This project is feasibility-first: the goal is not immediate rollout, but to measure potential benefits and risks, and to create an ever-evolving, plug-in platform that continuously adopts the best available open models for UK healthcare.
REFERENCES
– NHS England. A&E Attendances and Emergency Admissions – Statistical Commentary, March 2025.
URL: https://www.england.nhs.uk/statistics/statistical-work-areas/ae-waiting-times-and-activity/
– Royal College of Emergency Medicine (RCEM). Data show 1.65 million patients faced 12-hour waits from arrival in A&Es in 2022.
URL: https://rcem.ac.uk/press-release/data-show-1-65-million-patients-in-england-faced-12-hour-waits-from-time-of-arrival-in-aes-in-2022/
– Jones S, Moulton C, Swift S, et al. Association between delays to patient admission from the emergency department and all-cause 30-day mortality. Emergency Medicine Journal 2022;39:168-173
– Nuffield Trust – QualityWatch. Why is the planned care waiting list coming down? Analysis published 13 Aug 2025.
URL: https://www.nuffieldtrust.org.uk/news-item/why-is-the-planned-care-waiting-list-coming-down-and-what-does-the-data-really-tell-us
– Hanna TP, et al. Mortality due to cancer treatment delay: systematic review and meta-analysis. BMJ. 2020;371:m4087.
Direct URL: https://www.bmj.com/content/371/bmj.m4087
– Clinical Practice Research Datalink (CPRD). About CPRD. International Journal of Epidemiology Data Resource Profile. 2019.
URL: https://www.cprd.com
– UK Biobank. About our data.
URL: https://www.ukbiobank.ac.uk/about-our-data
– UK Biobank. World’s largest imaging project reaches 100,000 scans.
URL: https://www.ukbiobank.ac.uk/discoveries-and-impact/major-achievements/the-worlds-largest-imaging-project-reaches-milestone-of-100000-scans/