Use of AI-enabled stethoscope may help doctors detect more cases of Pregnancy-related heart failure, suggests study

Heart failure during pregnancy is a dangerous and often under-detected condition because common symptoms-shortness of breath, extreme fatigue and trouble breathing while lying down – are easily mistaken for typical pregnancy discomforts. Late-breaking research presented at the European Society of Cardiology Congress on a Mayo Clinic study showed an artificial intelligence (AI)-enabled digital stethoscope helped doctors identify twice as many cases of heart failure compared to a control group that received usual obstetric care and screening. Full study findings are published in Nature Medicine.

The trial was conducted in Nigeria, where more women experience pregnancy-related heart failure than anywhere in the world. The results also indicate that screening including the AI-enabled digital stethoscope were 12-times more likely than traditional screening to flag heart pump weakness when evaluated at a ejection fraction threshold lower than 45%, which is the cutoff indicating a specific type of heart failure called peripartum cardiomyopathy.

"Recognizing this type of heart failure early is important to the mother's health and well-being," says Demilade Adedinsewo, M.D., a cardiologist at Mayo Clinic and lead investigator of the study. "The symptoms of peripartum cardiomyopathy can get progressively worse as pregnancy advances, or more commonly following childbirth, and can endanger the mother's life if her heart becomes too weak. Medicines can help when the condition is identified but severe cases may require intensive care, a mechanical heart pump, or sometimes a heart transplant, if not controlled with medical therapy."

The randomized, controlled, open-label clinical trial included nearly 1,200 participants who were screened for heart conditions through typical obstetric care or AI-enhanced solutions. Mayo Clinic researchers previously developed a foundational 12-lead AI-electrocardiogram (ECG) algorithm to predict a weak heart pump, clinically known as low ejection fraction. A version of this algorithm was further enhanced by Eko Health for its point-of-care digital stethoscope which is U.S. Food and Drug Administration (FDA)-cleared to detect heart failure with low ejection fraction.

The researchers found that doctors using AI-based screening with the digital stethoscope and 12-lead ECG detected weak heart function with high accuracy. Within the study cohort, the digital stethoscope helped flag twice as many cases of low ejection fraction <50% and doctors using it were 12-times more likely to identify an ejection fraction <45% as compared to usual care.

The AI-supported tools were evaluated at three different levels of ejection fraction used in clinical diagnosis. Less than 45% is the cut point for diagnosing peripartum cardiomyopathy. Less than 40% indicates heart failure with reduced ejection fraction and has strong evidence in favor of specific medications to reduce symptoms and the risk of death. An ejection fraction of less than 35% signals severely low heart pump function that often requires more intense management, including advanced heart failure therapies and an implantable defibrillator if pump function does not recover. Patients in the intervention group each had an echocardiogram at study entry to provide confirmation of the AI-predictions.

"This study provides evidence that we can better detect peripartum cardiomyopathy among women in Nigeria. However, there are more questions to be answered," says Dr. Adedinsewo. "Our next steps would be to evaluate usability and adoption of this tool by Nigerian healthcare providers (including doctors and nurses) and importantly, its impact on patient care. Peripartum cardiomyopathy affects approximately 1 in 2,000 women within the U.S. and as many as 1 in 700 African American women. Evaluating this AI tool in the U.S. will further test its abilities in varied populations and healthcare settings."

Reference:

Adedinsewo, D.A., Morales-Lara, A.C., Afolabi, B.B. et al. Artificial intelligence guided screening for cardiomyopathies in an obstetric population: a pragmatic randomized clinical trial. Nat Med (2024). https://doi.org/10.1038/s41591-024-03243-9.