The back story to this story, Reid Thompson explains, is a longtime interest in sound, dating back to his undergrad days as a music major. It was only natural, he says, that his early training in medicine would lead him to pediatric cardiology and a focus on auscultation — listening with a stethoscope to detect heart-valve malfunctions.

“I was a musician, and what I learned I learned by listening as much as by looking,” says Thompson. “I liked using my ears.”

So much so that Thompson constructed what may be the world’s largest database of heart murmurs, what he calls “the ground truth for every sound the heart makes.”

That’s potentially critically important, Thompson explains, because for children, a heart murmur is one of the signs of heart disease. Children may be asymptomatic but have a murmur — a swishing sound in addition to the lub-DUB of a normal heartbeat as heart valves close. Literature reports that up to 70% of all children may have an innocent heart murmur at some point and do not need to be seen by a pediatric cardiologist and possibly undergo echocardiography, unlike the smaller number of children with pathologic murmurs who do need medical care.

However, Thompson notes, pediatricians may have a hard time distinguishing between the two due to a decline in auscultation skills. This prompted him to develop his database as a reference tool. In 2019, he went a step further and, in a pilot study in the Johns Hopkins ambulatory clinics, he used an electronic stethoscope to identify murmurs by telemedicine and artificial intelligence (AI). Patients’ murmurs could be sorted remotely into innocent and pathologic types. Inappropriate referral, he adds, often results in unnecessary echocardiography, emotional distress for parents and patients, time off from work and school, and labeling a child as having a potential heart problem despite cardiologists’ reassurances otherwise.

“Heart sounds were recorded at the primary point of care, transferred to a cloud-based server, listened to by a cardiologist and evaluated by AI following a paradigm that I thought would be a good way for pediatricians to refer patients who have a heart murmur,” says Thompson.

The study was placed on hold in early 2020 due to the arrival of COVID-19 and the restrictions on enrolling new patients in clinical areas. Then, Thompson had another idea, triggered by Johns Hopkins’ dramatic acceleration of its telemedicine program in response to the pandemic.

“We thought, what if we just send the electronic stethoscope to the home, where we could have the parents hold the stethoscope on the child’s chest during a video visit?” he says. “Up to now, all cardiologists could do during telemed visits was look and talk.”

How did the stethoscope-enhanced telemedicine visits go?

Thompson’s first patient was referred by a pediatrician for a suspected murmur. “I was impressed that we could hear a soft click sound, indicating an abnormal heart valve, very clearly remotely. Listening for a click is one of the hardest things to teach.” The patient was then scheduled for an echocardiogram (ultrasound of the heart), which showed the valve abnormality.

A visit with another patient Thompson had followed for mild pulmonary stenosis reassured him that the child’s murmur had not become louder and could wait to be seen. In another case, Thompson heard no murmur in an 11-month-old patient born with a hole in his heart, which meant the hole had closed and he did not need further testing. Another recent telemedicine patient, a 20-year-old with severe heart disease whom Thompson has followed for years, lives quite a distance from Johns Hopkins and was worried about leaving his home because of the coronavirus. Thompson noted that the shunt murmur was unchanged, and assured him he was fine and did not need to go to the hospital.

After the first dozen or so stethoscope-enhanced video visits, Thompson and his patients’ parents are sold.

“I can guide them in using the stethoscope, listen in real time, control the volume and filters, and the sounds are actually as good or better than if they were in clinic,” says Thompson. “The big news to me is this is going to work — the electronic stethoscope adds a huge dimension to just video. When I call parents and ask if they would be interested, they say, ‘Of course. Why wouldn’t we?’”

With electronic stethoscopes, a published study of AI algorithm accuracy on 600 patient heart recordings and institutional review board approval, Thompson is moving forward with more telemedicine visits. Pediatric cardiologists are asking him if the algorithm works, if the heart sounds are of diagnostic quality, if clinical decisions can be made this way. If so, will it change future practice regardless of pandemics?

Thompson answers, “This is not the whole solution. Maybe someday, we will be able to send them an echo probe — I don’t think that’s actually far-fetched. But in the meantime, and for patients who didn’t need that echo in the first place, this is a good step in using what we have now. Patients and parents attest to the fact that it was easy to do.”