‘Voice biomarker’ may predict chest pain, ACS, coronary artery disease
Click Here to Manage Email Alerts
WASHINGTON — A “voice biomarker” may be able to predict which patients have CAD or develop chest pain or ACS, according to findings presented at the American College of Cardiology Scientific Session.
For 108 patients (mean age, 59 years; 55% men; 57% with stable angina) clinically indicated for coronary angiography, researchers used an app (Vocalis Health) to make three 30-second voice recordings to establish a voice biomarker score, and calculated a voice biomarker value for each patient.
“This is not speech. This is the frequency of the voice that someone uses, like the frequency of an electrocardiogram,” Amir Lerman, MD, professor of medicine, consultant in the department of cardiovascular medicine, and director of the Chest Pain and Coronary Physiology Clinic at Mayo Clinic in Rochester, Minnesota, told Healio. “It’s just about the frequency that changes when you make sounds. We are able to analyze the domain of the frequencies.”
The researchers analyzed the relationship between voice biomarker score and the primary outcome of presenting to the ED with chest pain, being admitted for chest pain or having ACS, and the secondary outcome of a positive stress test result or the presence of CAD at follow-up angiography. Median follow-up was 24 months. The findings were simultaneously published in Mayo Clinic Proceedings.
“Several years ago, we got interested in the use of voice as a biomarker for identifying disease,” Lerman told Healio. “Because we are cardiologists, we wanted to see if it was related to coronary disease. Our first study showed that there are special domains in the voice that are characteristic of coronary disease. Our interest for this study was to see if that predicted events in coronary disease. As a physician, you do pay attention if a patient’s voice is different, even if it has not been an exact science.”
Patients were stratified by whether they had a high voice biomarker, defined as greater than 0.5217. The median voice biomarker level was 0.0675.
In a multivariable analysis adjusted for CAD grade on baseline angiography, patients who had a high voice biomarker level had elevated risk for the primary outcome (adjusted HR = 2.61; 95% CI, 1.42-4.8; P = .002) and the secondary outcome (aHR = 3.13; 95% CI, 1.13-8.68; P = .03) compared with those who did not, according to the researchers.
“The study showed that voice can be a noninvasive biomarker of disease, especially in these days of trying to collect data from the patient without the patient coming to the facility,” Lerman told Healio. “This information can be collected over the phone. Another message is that disease affects the body in different ways. If you have heart disease, it may not just be indicated by abnormalities in the heart. It can have a systemic effect.”
More research must be done before voice biomarkers can be used in clinical practice, Lerman told Healio. “We need larger and different populations to prove the concept beyond patients who come to Mayo Clinic,” he said. “One prior study showed that this is not language-dependent, which is important. But we need to show it works in other communities and populations. There is a signal, but we need to get more work done.”