Study confirms AI detection of HFpEF using a single echocardiogram view


Applying artificial intelligence (AI) to a single apical four chamber (A4C) view echocardiogram provides accurate information to detect heart failure with preserved ejection fraction (HFpEF), according to research published in JACC Advances.

The study, presented at the American Society of Echocardiography Annual Scientific Session (23–26 June, National Harbor, USA), demonstrates the platform could improve the diagnosis, management, and outcomes of a condition that currently often goes undetected, or requires additional invasive procedures to confirm, Ultromics said in a press release.

Ultromics’ EchoGo Heart Failure accurately detected HFpEF and provided fewer non-diagnostic outputs than current clinical scores, using just the routinely acquired A4C view from a transthoracic echocardiographic (TTE). The novel technology, which was recently granted clearance and Breakthrough Device designation from the US Food and Drug Administration (FDA), identifies radiomic signatures of disease that are not evident to the human eye.

Senior study author Patricia A Pellikka (Mayo Clinic, Rochester, USA), said: ”HFpEF can be difficult to detect, but left undetected and untreated, can result in hospitalization and mortality. As the first AI platform cleared to detect the condition, EchoGo Heart Failure can fill a significant unmet need.”

“With more than 32 million people living with HFpEF, and the incidence increasing, clinicians will benefit from having another means to recognise this disease.”

Based on the AI findings, patients could potentially be started on medications to treat their condition earlier than if they had to wait for an invasive diagnostic assessment of the disease.

The AI model was trained and developed on 6,756 patients who underwent a comprehensive TTE at Mayo Clinic in Rochester between January 2009 to December 2020. It was then independently tested in geographically distinct areas within Mayo Clinic enterprise System sites across the USA, on a dataset that included 1,284 patients.

EchoGo Heart Failure demonstrated high sensitivity and specificity, detecting 87.8% of patients who had HFpEF, and 81.9% of patients that did not. These results exceed what is usually observed in routine clinical practice.

It was also able to assign a correct diagnosis to 74% of patients who had returned non-diagnostic results on the commonly used HFA-PEFF and H2FPEF clinical scores. This improvement could translate to more patients receiving accurate and timely diagnoses and management.

During the follow-up period of up to five years, 444 patients died, highlighting the poor outcomes associated with HFpEF. The AI model was able to identify patients with worse survival, demonstrating its capacity to meaningfully improve patient outcomes.

Ross Upton, CEO and founder of Ultromics, said: “Our research demonstrates the tremendous potential of AI in revolutionising the detection of HFpEF. EchoGo Heart Failure’s exceptional discrimination capabilities combined with its ability to identify patients with higher mortality risks holds great promise for improving patient outcomes and enabling faster access to treatment.

“In a large number of cases, diagnostic data are often missing or discordant, making HFpEF detection challenging. AI can enhance echocardiography capabilities to help practices overcome the cumbersome intricacy of diastolic assessment. It is particularly useful for clinical centres that lack the time, resources, or expertise to perform comprehensive, diagnostic-quality, assessments.”


Please enter your comment!
Please enter your name here