Summary
Health Innovation Oxford and Thames Valley evaluated Ultromics’ AI-powered decision support tool EchoGo in stress echocardiography to predict the risk of coronary artery disease (CAD). Integrating this tool enhanced diagnostic accuracy and consistency, reduced subjective errors in interpreting data, and supported timely and informed clinical decisions. It led to more precise classification and better management of patients and showed potential for broader adoption within the NHS.
Problem being addressed
CAD is a major health concern. Reliable and rapid diagnostic methods are needed to guide effective treatment. Stress echocardiography is the established diagnostic, but it is subject to human error and variability in interpretation, which can impact diagnostic accuracy and treatment outcomes. Moreover, the growing volume of diagnostic data due to higher disease prevalence is straining current capabilities, making the case for integrating AI to improve consistency and reduce subjective errors in diagnostic processes. The EchoGo AI tool by Ultromics aims to address these challenges by automating the interpretation process, enhancing the objectivity and reliability of diagnostics in stress echocardiography.
Operational planning guidance/Government priorities
- Transition from analogue to digital: Supporting the NHS’s digital transformation initiatives to enhance healthcare delivery through advanced technology.
- Shift from treatment to prevention: Focusing on preventative measures and early diagnosis to manage health conditions effectively and reduce long-term healthcare costs.
Clinical area
Cardiology
What we did
We sought to establish whether AI improves the accuracy of stress echocardiography in predicting risk of coronary artery disease, improving patient outcomes and achieving cost savings. Our evaluation also sought to determine whether AI is a cost-effective solution for NHS-wide implementation.
A randomised controlled trial across 20 NHS hospitals with 2,213 patients was carried out, comparing stress echocardiography with and without AI assistance.
To assess sustainability sensitivity analysis was used to test different AI cost scenarios including installation, training, maintenance, diagnostic accuracy and clinician time.
A wide range of possible cost and outcome scenarios were modelled using Monte Carlo simulations to estimate cost-effectiveness of AI under different conditions.
What we found
AI-assisted stress echocardiography enhanced diagnostic accuracy, leading to more precise classification and better management of patients at risk of coronary artery disease, supporting timely and informed clinical decisions.
Use of the AI tool demonstrated cost-effectiveness – its economic value aligns with NICE affordability standards.
Potential time-saving for clinicians was identified, linked to streamlining workflows, reducing workload and enabling greater focus on more complex cases.
Patient outcomes were comparable both with and without AI input.
This diagnostic tool is a promising innovation with potential for broader adoption in the NHS.
Scalability/Next steps
Pilot programmes, training and real world evaluations should create an evidence base supporting adoption.
Contact: Ankur Chauhan, Senior Health Economist and Methodologist Ankur.chauhan@healthinnovationoxford.org