Point-of-care Heart Failure Diagnosis for GP Use: Implementation and Evaluation of a Simple AI-tool into the Folio of Care Pathways Serviced by Imperial’s Connected Care National GP Network

This first real-world evaluation of a simple-to-use AI-tool for GP point-of-care diagnosis of heart failure (HF) builds on the track record of the Connected Care Bureau (CCB) in establishing impactful AI-enhanced pathways across a growing Connected Care GP network. Consuming 2% of the entire NHS budget, 1M people in the UK have HF – increasing 50% by 2040.

Although 40% of patients will have seen their GP with symptoms of HF, the high demand, the importance of accuracy, and its difficulty results in diagnosis per NICE guidelines in only 4% of patients.

Consequently, 80% of HF is diagnosed only at life-threatening hospital admission, with 83% found to have the common form with reduced ejection fraction (HFrEF).

The Eko DUO digitally-enhanced stethoscope and cloud-based convolutional neural network is capable of determining if a patient has clinically significant HFrEF, at the national guideline-indicated treatment threshold of 40% EF with 100% sensitivity and 80% specificity.

The project will apply the Eko DUO in easily implemented point-of-care diagnostic pathways over 12 months in 10 GP practices across two UK regions in 200 patients in whom the GP considers HFrEF as a possibility.

All patients will also proceed through the established NHS pathway, with a repeat Eko DUO measurement performed for both direct correlation and reproducibility at the time of echocardiography - supplemented by real-world correlation with 300 unselected consecutive echos to explore refinements to diagnostic yield.

Accuracy of the Eko DUO, and the time to diagnosis and resource utilisation of the two pathways will be compared.

The twin challenges of early diagnosis of HF and its undiagnosed prevalence in the general population are among the biggest facing the NHS and would be transformed by an effective point-of-care diagnostic and pathway, anticipating our Phase 4 and 5 applications to the AI Award towards adoption nationally.