The goal of this clinical trial is to learn if the FIND HF algorithm detection rates of heart failure during testing are higher amongst participants identified as high FIND-HF risk compared to those identified as low risk in a population identified as at risk of undiagnosed heart failure. The main questions it aims to answer are:

• Are people who are identified as high risk by the FIND-HF tool more likely to be diagnosed with heart problems during testing than those identified as low risk?
• Are people identified as high risk by the FIND-HF tool more likely to show signs of heart problems during testing than those identified as low risk?? Researchers will compare patients in the high risk and low risk groups to see if more patients are detected with asymptomatic heart failure in the high risk group compared to the low risk group.

Participants will attend one visit at a local clinic where they will undergo an NT proBNP blood test which indicates heart failure and an echocardiogram to evaluate the heart's chambers, valves and overall function to help diagnose various heart conditions.

Background Heart failure (HF) affects around 1-2% of adults globally and is a leading cause of unplanned hospital admissions, especially among older people. In the UK alone, over one million people have HF, with about 200,000 new cases each year. This increase is mainly due to an aging population and better survival rates following heart attacks.

From the perspective of treatment, there are three main categories of HF:

• HF with reduced ejection fraction (HFrEF)
• HF with mid-range ejection fraction (HFmrEF)
• HF with preserved ejection fraction (HFpEF) For HFrEF, four pillar therapy (diuretics, beta-blockers, angiotensin-converting enzyme inhibitors, lanoxin) is established as the gold standard for pharmacotherapeutic management, helping to reduce the risk of death and hospital admissions. For HFpEF and HFmrEF, medications like empagliflozin, dapagliflozin, and finerenone have shown promising results in reducing risks. New drugs like Semaglutide and Tirzepatide are also beneficial for HFpEF patients with obesity. There are now more treatment options for HF than ever before.

In the UK, nurse-led teams in the community are capable of initiating and adjusting HF treatments. However, diagnosing HF early, before severe symptoms appear, remains challenging. Current guidelines recommend specific tests for diagnosis, but many (8 out of 10) diagnoses are still made in hospitals, often late.

The NHS aims to improve early diagnosis in primary care to reduce hospital admissions and improve patients' quality of life.

Funded by the British Heart Foundation (BHF) the investigators have developed an AI algorithm, FIND-HF, to identify people at risk of undiagnosed HF using data from routinely collected electronic health records.

Aims of the project The research project aims to determine the effectiveness of the FIND-HF AI algorithm in identifying HF at different risk levels and to understand the potential benefits of early detection and treatment.

The investigators will quantify the burden of symptoms, natriuretic peptide levels, and diagnostic yield of asymptomatic left ventricular systolic dysfunction, HFrEF, HFmrEF, and HFpEF at different FIND-HF AI risk scores. Using this information, the investigators will establish the risk score threshold that would be used for a future randomised clinical trial (RCT) of early detection of HF.

In addition, amongst patients newly diagnosed with HF during the study, the investigators seek to quantify the pharmacological treatment opportunities to inform the potential effect size, and thus sample size, of a future RCT of early detection and treatment of HF to reduce heart failure hospitalisation and cardiovascular death.

Identification of participants and recruitment Participants will be identified through by having their HF risk estimated their primary care electronic health records.

Individuals will be divided into two categories based on the predicted HF risk with recruitment 1:2 guided by risk strata, with 151 participants in the top 10 percentiles of risk, and 302 participants from the bottom 90 percentiles. This approach aims to enrich the cohort with higher risk patients while also including lower risk patients to determine whether there is an increase in the yield of new HF diagnosis from across the range of FIND-HF AI risk estimates.

Study invitations will be sent to eligible participants in each risk category in batches until the target sample size is reached. The invitation process will consist of a text message followed by an information pack in the post including a participant information sheet, consent form, contact information form and freepost return envelope. Participants will return the consent form and contact information to the research team who will then contact them to make an appointment for the study visit. The study visit will take place at the participant's GP surgery or a community-based health hub.

All participants will be required to provide written informed consent. Study visit

Consenting participants will attend one study visit. During the visit, a British Society of Echocardiography (BSE)-accredited clinical member of the research team will carry out the following tests/investigations:

• Medical history and symptom assessment (breathlessness, orthopnea, paroxysmal nocturnal dyspnea, exercise intolerance, fatigue, ankle swelling)
• Height, Weight, HR and BP
• Natriuretic peptide testing: Participants will undergo point of care (POC) natriuretic peptide testing
• Echocardiography: Participants will receive an AI-enabled hand-held echocardiogram using the Kosmos device and US2.ai automated measurements.

New cases of HF will be reported to the participant and their GP. Their GP will have responsibility to refer the participant as per local protocol.

Follow up Participants will be followed up remotely through their primary care EHRs at 6 months, and 1-, 5- and 10-years from their study visit by a member of the research team. Follow-up will include assessment of HF therapy and HF events. Consent will be sought for linkage to secondary care records, death certificate data and national registries.

Expected value of results The NHS Major Conditions Strategy stresses the need for evolving care models to better serve the public, with a focus on early diagnosis of cardiovascular disease using AI and digital technology. This proof-of-concept study will inform the design of a trial to test AI risk stratification and community-based pathways for early diagnosis and treatment of heart failure (HF), aiming to reduce HF events and cardiovascular death. If successful, the AI model (FIND-HF) could be implemented across UK primary care for a community-based early HF diagnostic pathway in the NHS.

Inclusion Criteria:

Registered with a doctor practice that uses electronic health system to record patient medical notes Age at enrolment greater or equal to 40 Identified as at risk of developing Heart Failure

Exclusion Criteria:

Registered with a doctor practice that is participating in research studies relating to heart failure screening

• Known diagnosis of HF
• On the palliative care register
• Unable to give written informed consent for participation in the study
• Unable to adhere to the study requirements

Patients identified at risk of developing Heart Failure by the FIND HF algorithm and are aged 40 and over, living in the Yorkshire area.

Planning to commercialise algorithm