The stethoscope has hung around doctors' necks for 210 years, and new research suggests the limiting factor is the ear at the other end of it. When five primary care clinicians listened to 357 at-risk patients with ordinary stethoscopes, they flagged 46.2 percent of the audible valve disease later confirmed by echocardiogram. Software analyzing the same hearts through a digital stethoscope flagged 92.3 percent.
Those findings, published in February in the European Heart Journal Digital Health, land as a coda to one of the more improbable origin stories in medicine. The instrument exists because a French physician in 1816 did not want to put his head on a young woman's chest.
The Awkward Consultation That Produced a Rolled-Up Sheet of Paper
René Théophile Hyacinthe Laennec was 35 and newly appointed to the Necker Hospital in Paris when a young woman arrived with what he called general symptoms of a diseased heart. Percussion, tapping the chest wall to judge what lies beneath, and laying a hand on it both offered little. Direct auscultation, pressing his ear to her skin, he ruled out anything related to her age, sex, and build.
So he reached for a piece of playground physics. Weeks earlier, in a courtyard at the Louver, he had watched two children signal along a length of solid wood: put your ear to one end and a pin scratching the far end is audible. Laennec rolled a sheaf of paper into a cylinder, set one end over the patient's heart and the other to his ear, and later wrote that he was surprised and elated to hear the beat far more clearly than his bare ear ever had.
He spent three years testing materials, settling on a hollow wooden tube roughly 25 centimeters long, and matching what he heard against autopsy findings. He named the technique mediate auscultation and the device the stethoscope, from the Greek for chest and to explore. His work gave medicine much of the vocabulary it still uses for chest sounds, including rales, egophony and pectoriloquy, and the first descriptions of bronchiectasis and cirrhosis. As a review of his life and work records, he died in 1826 at 45 of the tuberculosis he had spent his career characterizing with his own invention.
Human Ears Caught Fewer Than Half the Audible Valve Cases
The study was not a referendum on Laennec. It was a single-arm, single-blinded prospective trial run from June 2021 to May 2023 across three clinics, enrolling patients aged 50 and older with risk factors such as hypertension, diabetes or obesity but no prior valve diagnosis and no history of murmur. The median age was 70, and about 62 percent were women.
Every patient got two exams at four listening points. Five clinicians, four physicians and one nurse practitioner, used their own analog stethoscopes. Study coordinators separately recorded heart sounds on a digital stethoscope for analysis by an FDA-cleared suite of convolutional neural network algorithms. Echocardiography plus an independent expert panel set the ground truth, and both were blinded to the algorithm and echocardiogram results.
Against audible moderate or worse valve disease, sensitivity more than doubled. Against every echocardiographically confirmed case, including murmurs no human could reasonably hear, it rose from 13.8 to 39.7 percent. The algorithms identified 12 patients with previously undiagnosed clinically significant valve disease. Routine listening found six.
The authors point to earlier work in the journal Heart finding roughly 44 percent sensitivity in asymptomatic patients when experienced general practitioners listened for significant valve disease.
The Trade-Off Buried in the Doubled Detection Rate
The obvious question is what the software costs. Specificity fell, from 95.6 to 86.9 percent on the audible-murmur comparison and from 94.7 to 88.6 percent against echocardiographic disease. More false positives mean more normal echocardiograms.
The authors are direct about that tension and about a second limitation: earlier detection is not the same thing as better outcomes. They write that finding disease sooner may help prevent irreversible cardiac damage when followed by appropriate management, and they call for cost-effectiveness studies that have not been done. Six of the nine authors are current or former employees of Eko Health, which makes the device, a conflict disclosed in the paper. The work was supported by a National Science Foundation grant.
The study was also small, run in clinics within one geographic region, and did not record whether patients had symptoms at enrollment. Undiagnosed valve disease of some kind is present in more than half of adults over 65, so the population at stake is large. A single 357-patient study does not settle how to screen it.
Veterinary Clinicians Just Beat the Same Class of Technology
Earlier this month a different result cut the other way, involving a stethoscope from the same manufacturer. Researchers at North Carolina State University tested an Eko digital stethoscope on 105 companion animals, 54 dogs and 51 cats, enrolled between August and December 2025. Each was examined by a board-certified cardiologist, a cardiology resident, and a fourth-year veterinary student.
In dogs, the device detected murmurs about as well as the students did, at 86.8 percent sensitivity. In cats, it collapsed, finding only 2 of 22 murmurs while students caught nearly two-thirds. Arrhythmia calls were unreliable in both species: the device never once labeled a dog's rhythm normal. The findings appeared in the Journal of the American Veterinary Medical Association and were summarized by NC State, which noted that veterinarians use the same stethoscopes human physicians do, and that the algorithms are trained on human data.
That is the more useful frame for both, and close to how the European Society of Cardiology framed the human result: the device requires clinical judgment rather than replacing it. The algorithm performed well on the population it was built for, and poorly outside it. Laennec's insight was that a medium between patient and physician could carry information the naked ear could not. Two centuries later, the argument is about what belongs at the listening end.
Key Questions Answered
Who actually invented the stethoscope and why? René Laennec, a French physician, in 1816 at the Necker Hospital in Paris. He rolled paper into a tube because he considered it inappropriate to press his ear directly against a young female patient's chest.
What did the new study find? In 357 primary care patients aged 50 and older, clinicians using ordinary stethoscopes detected 46.2 percent of audible moderate-to-severe valve disease. Software analyzing digital recordings detected 92.3 percent.
Does that mean my doctor's stethoscope exam is useless? No. Auscultation remains a fast, free screening step, and it caught six previously undiagnosed cases in this study. The finding is that it misses a substantial share of significant valve disease, particularly in patients without symptoms.
What is the downside of the software version? More false alarms. Specificity dropped by roughly nine percentage points, which would mean more patients sent for echocardiograms that come back normal.
Was the study independent? Partly. Six of the nine authors are current or former employees of the company that manufactures the digital stethoscope, and the paper discloses this. The work was supported by a National Science Foundation grant.
What should patients do with this information? Nothing dramatic. Anyone with unexplained breathlessness, chest discomfort, fainting, or reduced exercise tolerance should raise it with a clinician rather than assume a normal exam ruled out valve disease.
