AI-enhanced heart failure screening: Is it worth it?

According to a Mayo Clinic report, incremental drops in heart function are treatable with medication, although they tend to be difficult to spot. This is because there is no guarantee that patients will show symptoms of their hearts pumping ineffectively. Without symptoms, there’s no guarantee that physicians will even order an echocardiogram or other diagnostic tests.

A study published in Nature Medicine in May 2021 demonstrated that primary care clinicians using artificial intelligence (AI) tools to conduct electrocardiograms (ECGs) were able to effectively screen patients for signs of heart failure, successfully identifying more unknown cases of a weak heart pump (also known as a low ejection fraction), than they were able to without the use of AI. New research, published in the Mayo Clinic Proceedings: Digital Health, looked to uncover whether the use of AI-ECG tools was good value for the money spent by practices. Their findings suggest that AI-enhanced heart failure screening is cost-effective in the long term, especially in outpatient settings.

When compared with usual care, AI-ECG was considered to be cost-effective, with an incremental cost-effectiveness ratio (ICER) of $27,858 per quality-adjusted life year (QALY)—a measurement combining a patient’s life expectancy with their quality of life. Researchers reported that the program remained cost-effective even with a change in patient age and follow-up time duration, although, they noted, the specific ICER values varied for these parameters. The program was found to be particularly cost-effective in outpatient settings than in emergency department settings, with a significantly lower ICER of $1,651 per QALY.

The study analyzed the cost-effectiveness of the AI-ECG tool through the use of real-world information from 22,000 participants in the established 2021 EAGLE trial, following which patients had weak heart pumps and which did not. Researchers conducted a simulation into the progression of the later stages of the disease, assigning values for the health burden on patients and the effect that had on economic value.

“We categorized patients as either AI-ECG positive, meaning we would recommend further testing for low ejection fraction, or AI-ECG negative with no further tests needed,” Xiaoxi Yao, PhD, MPH, senior author of the study and professor of health services research at Mayo Clinic, said in an organizational release. “Then, we followed the normal path of care and looked at what that would cost. Did they have an echocardiogram? Did they stay healthy or develop heart failure later and need hospitalization? We considered different scenarios, costs and patient outcomes.”

Yao emphasized the significance of cost-effectiveness in the evaluation of AI tools and technologies when considering their implementation into clinical practice. “We know that earlier diagnosis can lead to better and more cost-effective treatment options,” he said. “To get there, we have been establishing a framework for AI evaluation and implementation. The next step is finding ways to streamline this process so we can reduce the time and resources required for such rigorous evaluation.”