AHA invests in artificial intelligence to address gender bias in cardiology
Over the past decade, healthcare leaders have become increasingly aware of the fact that cardiology has a significant gender bias problem.
Heart disease is the leading cause of death in women, yet it remains widely underrecognized, misdiagnosed and undertreated. Women are more likely than men to present with atypical symptoms and are less likely to be referred for diagnostic testing or aggressive treatment, in part because research has historically focused on male patients. As a result, women often experience delays in care and worse outcomes.
Last week, the American Heart Association invested in a British startup selling artificial intelligence models to help close the diagnostic gap for heart failure. The investment is made through the American Heart Association's Go Red for Women Venture Fund, which was launched last year to help address long-standing gender disparities in the diagnosis and treatment of heart disease.
The startup, called Ultromics, was founded in 2017 by Ross Upton, who currently serves as its CEO and chief scientific officer. He said the company was based on work he did while earning his PhD in cardiovascular medicine at the University of Oxford, where he trained as a cardiac sonographer and saw firsthand the subjectivity of echocardiogram interpretation and how easily diagnoses could be missed.
“Many cardiovascular diseases go undiagnosed because patients are not tested,” Upton declared. “They go undiagnosed because the large-scale testing we have done does not consistently provide clinicians with clear answers for certain conditions.”
He noted that the two most important examples are cardiac amyloidosis and heart failure with preserved ejection fraction (HFpEF).
Upton explained that these conditions are difficult to identify early with routine echocardiography, even for experienced clinicians, and they often present with symptoms that are difficult to describe. This can result in symptoms persisting, test results remaining inconclusive, and treatment starting later than it should.
“This disparity disproportionately affects women with HFpEF. Symptoms such as fatigue, shortness of breath, and swelling are more likely to be ignored or attributed to noncardiac causes,” Upton said.
He added that studies show that women are more likely to develop HFpEF than men and that up to 64% of cases go undiagnosed in clinical practice.
Ultromics' technology aims to identify HFpEF earlier and more consistently by using artificial intelligence to analyze echocardiograms. Rather than relying solely on what people can detect visually when quickly viewing an echocardiogram video, the model examines patterns of motion throughout the clip.
“Technically, it's analyzing millions of pixels from multiple heartbeats and seeing if those patterns match patterns we see in patients with known diseases,” Upton said.
The platform is currently in use at health systems including Northwestern Medicine, University Hospitals and University of Chicago Medicine. To date, Ultromics has processed more than 430,000 echocardiograms worldwide.
HeartFlow offers similar technology, but Upton doesn't consider the company a direct competitor. He noted that its platform is similar in that it takes a standard, widely used cardiac imaging test and extracts additional clinically useful signals from it without requiring hospitals to change how the scans are acquired.
Ultromics seeks to achieve this goal as well—but in a different way and disease area.
“We work with echocardiography, which typically has much higher volumes than CT, and we return results in minutes rather than hours, so the output naturally fits into routine workflows,” Upton declared.
Ultromics is focused on hard-to-diagnose diseases, specifically HFpEF and cardiac amyloidosis, where early detection can change a patient's trajectory, he added.
He said the company will use the new funding from the AHA to expand adoption of its platform in U.S. hospitals, aiming to address some of the most persistent blind spots in cardiology.
Photo: Jaime Grajales Benjumea, Getty Images
