According to research data shared at the American Heart Rhythm Society meeting this week, researchers at the Mayo Clinic have created an artificial intelligence algorithm that can use a single-lead electrocardiogram taken by an Apple Watch to find patients with low effective ventricle pumping.
Low effective ventricular pumping or left ventricular dysfunction is a problem that affects 2 to 3 percent of people worldwide, and up to 9 percent of people over the age of 60. As with atrial fibrillation, another heart problem that the Apple Watch can detect, low effective pumping of the ventricle can be asymptomatic. It may also be accompanied by some symptoms, including a fast heartbeat or shortness of breath.
Paul Friedman, chair of the Mayo Clinic’s Department of Cardiovascular Medicine, said the ability of AI to detect the condition using the ECG capabilities of consumer smartwatches is “absolutely remarkable,” since it typically requires an echocardiogram, CT scan or MRI to identify.
The ECG feature on the Apple Watch is a single-lead ECG that requires the user to place their finger on the Apple Watch’s digital crown for 30 seconds. Results are uploaded to the Apple Health app and can be shared with medical professionals. The ECG is designed to help detect atrial fibrillation, but the ECG feature and other features of the Apple Watch are also being investigated for use in detecting other conditions.
A standard ECG uses 12 electrode leads, placed on a person’s chest, arms, and legs, to monitor electrical signals from the heart. To use the Apple Watch’s single-lead ECG results, the researchers modified an existing 12-lead algorithm known to detect weakness in the heart muscle.
The study included 125,610 ECGs collected from 46 states and 11 countries over a six-month period. Each person submitted multiple ECG reports, and the cleanest readings were used for the algorithm. Several hundred participants underwent clinical tests to measure the strength of the pump, and the data was used to determine whether the Apple Watch could detect problems.
About 420 patients had a watch ECG recorded within 30 days of a clinically ordered echocardiogram or cardiac ultrasound, a standard test for measuring pump strength. We used this data to see if we could identify conditions where the effective pumping force of the ventricle is low through AI analysis of the watch’s ECG. Although our data is early, the area under the curve for this test is 0.88, which means it is as good or slightly better than the medical treadmill test. AI analysis of the watch’s ECG is a powerful test that can identify low effective pumping of the ventricle.
The researchers plan to initiate a global prospective study with further testing in more diverse populations to demonstrate the benefits of the single-lead ECG feature in the Apple Watch.
“This is what a transformation in medicine looks like: diagnosing serious diseases cheaply from your couch,” Friedman said.