This 15-Year-Old Developed Machine Learning Tech to Screen for Cardiac Risk

Over 320,000 cardiac arrests occur annually in the U.S. outside a hospital, making it one of the leading causes of death in the country.

Quickly and without warning, an individual can lose consciousness in seconds as the heart’s pumping action is disrupted.

Cardiac arrest is also the leading cause of death in athletes. But sports physicians often have a difficult time determining cardiac risk in their patients.

At the age of 10, Sofia Tomov witnessed her grandmother suddenly collapse from a heart attack. While she recovered, no one in the family previously had any idea that she had heart disease.

“That was something that struck me as terrifying — how someone can seem fine one moment and then collapse the next,” now-15-year-old Tomov tells Hypepotamus. “That’s when the idea for my business really hit me.”

She reviewed existing scientific literature and spoke with several cardiac experts and cardiologists during her time as a dual enrollment high school student at University of Tennessee — Knoxville.

After taking a class on machine learning, she saw how she could apply the technology to predict heart disease risk.

“Researchers have done some work for predicting heart disease risk, but there’s still a compelling need to improve accuracy and develop a technology that could be commercialized to help meet a doctor’s need directly,” she says.

Tomov founded Qardian Labs to bring her heart disease prediction solution to market. The HIPAA-compliant software fuels its algorithm with 11 patient data metrics to determine heart disease risk in only a few seconds. The metrics include blood sugar, cholesterol, age, sex, maximum heart rate achieved, and other physical test results.

“The algorithm uses a deep neural framework to learn relationships between the 11 patient data points,” says Tomov. “That’s particularly significant since most doctor screening tools only analyze numbers one-by-one.”

The dashboard displays the patient’s heart disease probability risk, which the doctor can use to develop a treatment plan.

The startup, which works for both regular patients and athletes, is undergoing a pilot in the University of Tennessee Medical Center’s sports medicine practice to use echocardiogram data as a standard screening procedure for student athletes.

The pilots will continue until the end of the year. Once launched, Qardian Labs will operate on a licensing revenue model for health systems and individual medical practices.

Though evaluating investor partnerships, thus far Tomov has raised only non-dilutive capital for Qardian Labs, largely through pitch competitions. Last month, she took home $10,000 from the the 36|86 Student Pitch Competition.