Voice Biomarker Can Predict Coronary Artery Disease Events

A research team led by Mayo Clinic created an artificial intelligence (AI) computer-based algorithm that can use voice biomarkers to locate clogged arteries, helping clinicians detect heart conditions.

The study published in Mayo Clinic Proceedings included 108 participants who participated in three 30-second voice recordings using a smartphone app called Vocalis Health. All recordings took place between Jan. 1, 2015, and Feb. 28, 2017.

The app leveraged an algorithm that analyzed participants' voice samples, identifying more than 80 voice recording features, such as frequency, amplitude, pitch, and cadence. Researchers took six features highly correlated with coronary artery disease (CAD) and combined them to form a single score. One-third of patients in the study received a high score and two-thirds a low score.

Researchers found that when the score was high, it resulted in a high hazard ratio. A high noninvasive voice biomarker score was linked to a higher risk of a CAD-related medical event.

Specifically, 58.3 percent of patients with a high voice biomarker score went to the hospital due to chest pain or acute coronary syndrome. This is significantly higher than the 30.6 percent of low voice biomarker score patients who went to the hospital.

“We can’t hear these particular features ourselves,” said Jaskanwal Deep Singh Sara, MD, a cardiology fellow at Mayo Clinic and the study’s lead author, in the press release. “This technology is using machine learning to quantify something that isn’t easily quantifiable for us using our human brains and our human ears.”

Researchers believe that the autonomic nervous system is responsible for connecting the voice and CAD. The autonomic nervous system controls subconscious bodily functions, including the vocal cords and cardiovascular system.

Despite the potential connection between CAD events and a noninvasive voice biomarker, researchers admit that more research is needed to confirm their findings, as results could differ in various healthcare settings, countries, and demographics.

“It’s definitely an exciting field, but there’s still a lot of work to be done,” said Sara. “We have to know the limitations of the data we have, and we need to conduct more studies in more diverse populations, larger trials and more prospective studies like this one.”

The study results will be presented at the American College of Cardiology’s 71st Annual Scientific Session in April.

Another recent study described a method of detecting heart conditions using artificial intelligence (AI). Two frequently overlooked conditions called hypertrophic cardiomyopathy and cardiac amyloidosis can lead to further issues for patients. However, an AI algorithm detected abnormal activity such as the thickening of the heart muscle and the building up of an unhealthy protein called amyloid, pinpointing the heart conditions.

Another study published in December 2021 describes an AI model that monitors metabolic and cardiovascular biomarkers, which can predict expected lifespan based on cardiovascular health. The AI model used seven metabolic biomarkers, including body mass index, cholesterol, and triglycerides, and determined how they correlate with aging.

Further, the voice biomarker study results were released as Google announced plans to research whether the built-in microphones on a smartphone can record heart sounds when placed on the chest and help detect heart valve disorders.