AI analysis of smartwatch data detects hypertrophic cardiomyopathy with 98% accuracy

Using a machine learning algorithm to analyze blood volume data gathered by commercial smartwatches is a reliable screening system for hypertrophic cardiomyopathy (HCM), according to a study published June 24 in npj Digital Medicine.

In the study, scientists from South San Francisco-based precision cardiovascular medicine company MyoKardia incorporated noninvasive optical sensors into smartwatches to gather photoplethysmography recordings and echocardiograms from HCM patients. The data was then fed to MyoKardia's proprietary machine learning algorithm.

As a result, the algorithm correctly identified 95 percent of patients with HCM, 18 of the 19 studied. Additionally, the algorithm correctly found 63 of the 64 members of the control group to be healthy, for an overall operating accuracy of 98 percent.

Per the study's authors, this method therefore presents a noninvasive and cost-effective way to screen for HCM. The condition is associated with an increased risk of heart failure, stroke and sudden death and is "vastly underdiagnosed," with only an estimated 15 percent of HCM patients receiving the proper diagnosis, according to senior study author Marc Semigran, MD, senior vice president of medical sciences at MyoKardia.

More articles about AI:
1st noninvasive, brain-controlled robotic arm holds potential for stroke, paralysis treatment
FDA clears tool for radiologists to accept, reject, interact with AI imaging data
Updated White House AI policy prioritizes public-private collaboration

© Copyright ASC COMMUNICATIONS 2021. Interested in LINKING to or REPRINTING this content? View our policies by clicking here.