AI stethoscope doubles detection of severe valvular heart disease in primary care research

Emerging evidence suggests that AI-assisted auscultation can help clinicians detect hidden heart valve disease early and has the potential to reshape front-line cardiac screening, while raising important questions about the balance between implementation and diagnosis.

Research: Artificial intelligence-enabled digital stethoscope improves point-of-care screening for moderate to severe valvular heart disease. Image credit: Natali _ Mis / Shutterstock

In a recent prospective study published in European Heart Journal Digital HealthResearchers compared the diagnostic accuracy of primary care providers using a standard stethoscope with that of a relatively new artificial intelligence (A.I.) Compatible digital stethoscope. This study aimed to determine whether the latter could improve the accuracy of current diagnostics of heart valvular disease (VHD).

As a result of the research, A.I. The system demonstrated a sensitivity of 92.3% for detecting audible sounds. VHDcompared with 46.2% with standard care (P = 0.01). However, A.I. Although the tool showed slightly lower specificity, it identified twice as many cases of previously undiagnosed moderate-to-severe disease, suggesting a role as an adjunct to screening rather than a substitute for clinical evaluation.

background

Valvular heart disease is a serious heart disease in which one or more heart valves, such as the aortic, mitral, tricuspid, or pulmonic valves, do not open and close properly, blocking blood flow.

Common symptoms include shortness of breath, fatigue, chest pain, and palpitations. The prevalence of the disease increases with age, with estimates that more than half of adults over the age of 65 are affected to some extent, although moderate to severe disease is much less frequent.

Diagnosis remains difficult because more than half of patients with clinically significant disease are asymptomatic.

Traditionally, diagnosis has relied on auscultation performed by a clinician. However, previous studies suggest that screening asymptomatic patients, even for experienced general practitioners, has limited sensitivity, contributing to delays in diagnosis and disease progression.

Research design and methods

This study investigated whether deep learning algorithms combined with digital acoustic recordings could detect heart abnormalities that may be missed by routine tests.

This was a prospective single-arm diagnostic accuracy study conducted in three primary care clinics between June 2021 and May 2023. The cohort included 357 patients aged 50 years and older who were at high cardiovascular risk but had no previous diagnosis of cardiovascular disease. VHD or a known heart murmur.

Risk factors include high blood pressure and body mass index (BMI).BMI) >30, diabetes, hyperlipidemia, atrial fibrillation, history of myocardial infarction, stroke or transient ischemic attack, coronary revascularization, or other established cardiovascular disease.

Participants underwent two independent screening protocols.

Standard treatment (SOC) screening, primary care provider (PCP) Four-point cardiac auscultation was performed using a conventional stethoscope.

in A.I.– Extended screening, study coordinator records phonocardiogram (PCG) Acquire data using a digital stethoscope. The recordings were analyzed in the following manner. A.I. algorithm cleared by F.D.A. Detect heart murmur.

All participants underwent echocardiography to confirm structural heart disease. An independent panel of experts reviewed digital audio recordings to confirm the existence of audible tweets, but was not informed. A.I. result.

audible VHD Recognizing that some structurally significant disease may not produce a clearly audible murmur, it was defined as moderate to severe disease confirmed by echocardiography with an audible murmur confirmed by an expert.

Research results

of A.I.– The augmented system significantly outperforms standard auscultation in detecting audible sounds. VHD. The sensitivity was 92.3%. A.I. compared to 46.2% of SOC Screening (P = 0.01).

Of the confirmed cases, 7 out of 13 patients were missed by standard tests; A.I. The system only missed one. For previously undiagnosed moderate to severe patients VHD, A.I. 12 cases were identified; PCPs.

This increase in sensitivity was accompanied by a decrease in specificity. of A.I. The specificity of the system was 86.9 percent versus 95.6 percent for clinicians (P < 0.001), resulting in more false-positive findings.

Using echocardiography alone as the reference standard for moderate to severe disease, regardless of the audibility of the heart murmur, A.I. The system still outperformed standard of care, with sensitivity of 13.8 percent compared to 39.7 percent for clinicians (P = 0.01).

conclusion

By integrating this research, A.I.– Introducing enabled digital stethoscopes into primary care could significantly improve symptom detection. VHD compared to traditional auscultation. These tools serve as a second layer of screening support and allow for early identification and referral.

Because this study evaluated diagnostic accuracy rather than downstream management or prognosis, early detection does not automatically lead to improved clinical outcomes.

Despite disclosing conflicts of interest, several authors reported that relationships with device manufacturers should be considered when interpreting study results.

Low specificity may increase echocardiography referrals and health care utilization, highlighting the need for future cost-effectiveness analyses.

Limitations include modest sample size, limited geographic coverage, incomplete demographic details, and lack of systematic symptom assessment. Despite these limitations, our findings show that: A.I. Augmentation could lead to meaningful advances in point-of-care cardiac screening.