Digital Tech, A.I. Assessed for Heart Valve Problems

(Gerd Altmann, Pixabay)

8 Mar. 2019. Digital stethoscopes combined with artificial intelligence will be evaluated in a clinical study of these technologies to screen for heart valve disease. Northwestern University’s medical center in Chicago will assess the ability of digital stethoscopes made by Eko Devices Inc. in Berkeley, California, with results analyzed by algorithms to accurately detect and interpret heart murmurs as signs of heart valve disorders.

The human heart has 4 valves — mitral, tricuspid, pulmonary, and aortic valves — that keep blood flowing in the proper direction. The valves, when operating correctly, open and close once during each heart beat. When heart valves don’t function properly, blood flow is disrupted, leading to possible complications such as heart failure, stroke, blood clots, or irregular heart rhythms. These valve disruptions can occur due to valves not closing properly, or thickening and hardening of the valves that narrow the opening and restrict blood flow. Heart valve problems can also be congenital, where the valves do not form correctly or at all in babies during pregnancy.

Doctors check for heart valve disorders during routine checkups using their stethoscopes, perhaps the one piece of medical equipment most familiar to patients. While considered a trusted tool of medicine, using a stethoscope correctly is as much an art as a science, says Eko, requiring a trained ear to separate normal heart functions from murmurs that can be signs of diseased heart valves. And while more advanced equipment like echocardiograms that use ultrasound to form a real-time image of a working heart, that level of sophistication is not always available in many clinics and doctors’ offices for simple heart disease screening.

Northwestern Medicine’s cardiovascular institute is assessing digital stethoscopes made by Eko to provide a more reliable diagnosis of heart valve difficulties. Eko makes a handheld device called the Duo that combines a stethoscope and single-lead electrocardiogram or ECG. The Duo has an earpiece and connects wirelessly to a mobile app that interprets and displays the captured signals. Eko also makes a more traditional stethoscope that combines conventional acoustic with electronic functions.

The study is also evaluating machine learning algorithms for interpreting the data generated by digital stethoscopes to detect heart valve disease. In June 2018, Northwestern Medicine received a $25 million gift from the Bluhm Family Charitable Foundation that endows the medical center’s cardiovascular institute. That gift includes funds for a center to study the power of artificial intelligence to improve diagnosis, treatment, and research for cardiovascular disease.

The project includes a clinical trial funded by Eko enrolling 1,000 individuals, most of which at Northwestern Medicine. Eko also has a clinical study now underway to train a machine learning algorithm to detect heart valve problems with 900 participants at University of California in San Francisco. In January, a similar study published in Nature Medicine (paid subscription required) shows an algorithm trained by ECG data from some 45,000 patients at the Mayo Clinic could accurately and reliably detect low ejection fraction, a form of heart failure marked by a malfunctioning left heart ventricle, often without perceivable symptoms.

“If proven effective, Eko’s platform could be a much simpler, lower cost way to identify patients with heart disease,” says Northwestern cardiologist James Thomas, the principal investigator of the project in an Eko Devices statement. “We are looking to support and advance work that broadens access to the best diagnostic tools in health care, regardless of whether a patient lives in the city or a more rural area. Deep learning provides that expert knowledge, regardless of a patient’s location.”

More from Science & Enterprise:

• High-Speed Screens, A.I., Boost Nanoscale Medicines
• Smart Watch in Development for Respiratory Diseases
• Algorithms Identify More Genetic Syndromes
• AI, Wearables Posing Health Privacy Risks
• Validation Scheme Proposed for Medical Algorithms

*     *     *