NIH to Help Fund Development of ML Algorithm for Cardiovascular Care

With financial support from the National Institutes of Health (NIH) and the Department of Health and Human Services (HHS), Eko aims to create a machine-learning algorithm for cardiovascular care that can detect conditions such as pulmonary hypertension (PH).

Following the receipt of a $2.7 million Small Business Innovation Research (SBIR) Direct Phase II grant from the NIH, which is part of HHS, Eko plans to develop a machine-learning algorithm that can use phonocardiogram (PCG) and electrocardiogram (ECG) data collected by the company's smart stethoscopes to identify and stratify PH.

Eko will use PCG and ECG data gathered through the Eko DUO ECG + Digital Stethoscope within the Lifespan Health System Cardiovascular Institute. Through this collaboration, Eko developers aimed to collect data that would ultimately assist in creating and testing an algorithm that could detect PH at a fast pace.

Affecting up to 1 percent of the global population, PH involves the buildup of pressure in vessels that carry blood to the heart and the lungs. Those with PH may experience various adverse health outcomes, including heart failure. In some cases, the period between the start of symptoms and diagnosis of PH can be as long as two years.

"The major goal of this study is to determine whether an Eko algorithm based on phonocardiography coupled with electrocardiography can identify the presence and severity of pulmonary hypertension when compared to the current gold standard," said Gaurav Choudhary, MBBS, principal investigator, and Ruth and Paul Levinger Professor of Cardiology and director of Cardiovascular Research at the Alpert Medical School of Brown University and Lifespan Cardiovascular Institute, in the press release.

Choudhary also pointed out various benefits that could accompany the use of the algorithm.

"This machine-learning algorithm has the potential to be a low cost, easily implementable, and sustainable medical technology that assists healthcare professionals in identifying more patients with pulmonary hypertension,” he continued in the press release.

The use of machine learning and artificial intelligence to detect and diagnose conditions is becoming an increasingly common practice.

A study published in the American Journal of Roentgenology in August provided insights into the ability of machine-learning models to predict tumor recurrence among patients with hepatocellular carcinoma.

Another study published in February found that machine learning could provide details regarding the severity levels of COVID-19. The study included testing three different machine-learning techniques, allowing providers to understand which was the most accurate and how they can be used in potential future interventions.