AI-ECG Tools Can Identify Heart Muscle Weakness in Women Before Pregnancy

Each year, some mothers die from heart-related issues after childbirth, with many of these deaths being preventable. Screening for heart weakness before pregnancy could be crucial in identifying women who may require additional care to improve pregnancy outcomes. Now, researchers have tested artificial intelligence (AI) tools, using electrocardiogram (ECG) recordings and digital stethoscopes, to detect undiagnosed heart conditions in women of childbearing age attending primary care. The study, published in Annals of Family Medicine, demonstrates the high diagnostic performance of these technologies in identifying left ventricular ejection fraction below 50%, a sign of heart muscle weakness.

In this cross-sectional pilot study, researchers from Mayo Clinic (Rochester, MN, USA) evaluated the effectiveness of an AI-enabled ECG and an AI-powered digital stethoscope in two groups of women aged 18 to 49 who were considering pregnancy. The study also included women who were pregnant or within one year postpartum. The first group consisted of 100 women who were already scheduled for an echocardiogram, while the second group included women who had no indication for an echocardiogram but were attending a primary care appointment for routine checkups. All participants underwent two tests: a standard 10-second 12-lead ECG and a digital stethoscope recording, which captured a 15-second, single-lead ECG and phonocardiogram (heart sounds) from up to three chest locations. AI models then analyzed the ECG and stethoscope recordings to assess each participant's risk of left ventricular systolic dysfunction (LVSD), a type of heart dysfunction.

The AI-ECG showed an area under the curve (AUC) of 0.94, while the AI-powered digital stethoscope, Eko DUO, achieved an even higher AUC of 0.98, indicating strong diagnostic accuracy. In the second cohort, the prevalence of positive AI screening results was 1% for the AI-ECG and 3.2% for the AI-stethoscope. This research builds on earlier studies, including a pilot prospective study evaluating AI digital tools to detect pregnancy-related cardiomyopathy among U.S. obstetric patients, and a pragmatic randomized clinical trial conducted with women in Nigeria who were pregnant or had recently given birth. Together, these findings highlight the potential of AI to transform cardiovascular screening, enabling earlier identification and management of heart muscle weakness in women of reproductive age. Ongoing research aims to explore the potential of these technologies for screening heart weakness in broader populations.

"Statistically, nearly half of pregnancies in this country are unplanned, and approximately 1% to 2% of women may have heart problems they don't know about,” said Demilade Adedinsewo, M.D., a cardiologist and senior author of the study. “Our research findings suggest that these AI tools could be used to screen women before pregnancy, allowing for improved pregnancy planning and risk stratification, early treatment, and better health outcomes which addresses a critical gap in current maternal care."