The goal of this observational diagnostic study is to evaluate whether an artificial intelligence (AI)-enabled smart stethoscope can accurately detect structural heart disease in school-aged children and adolescents (10-18 years) in Ruyang County, China.

The main questions it aims to answer are:

Can the smart stethoscope reliably identify students with cardiac murmurs that indicate possible structural heart disease? How well do the sensitivity, specificity, and predictive values of the smart stethoscope compare with standard echocardiography?

Researchers will compare AI-assisted stethoscope screening results with echocardiography (gold standard) to see if the device can be used as an effective early screening tool.

Participants will:

Undergo a heart sound screening using the AI-enabled smart stethoscope (3-5 minutes).

If screening is positive, receive a free echocardiogram at Ruyang County People's Hospital.

A small sample of students with negative screening results will also receive echocardiography to check for missed cases.

Structural heart disease (SHD), including congenital and acquired cardiac abnormalities, is a leading cause of morbidity in children and adolescents. Cardiac murmurs are common clinical signs, but traditional auscultation has limited accuracy in school or community settings due to examiner variability and limited access to echocardiography.

This study evaluates the performance of an artificial intelligence (AI)-enabled smart stethoscope for school-based screening of SHD in primary and secondary students in Ruyang County, China. The device integrates high-sensitivity acoustic sensors, noise-reduction technology, and deep learning algorithms to provide automated interpretations of heart sounds within seconds. Prior validation studies have demonstrated high sensitivity (>80%) and specificity (>90%) for congenital heart disease and up to 94% sensitivity and 98% specificity for rheumatic heart disease.

Screening will be conducted by trained personnel at four standard cardiac auscultation sites. Students with abnormal AI findings will undergo repeat testing and, if confirmed, will be referred for transthoracic echocardiography at Ruyang County People's Hospital. A subset of students with negative AI screens will also receive echocardiography to estimate false-negative rates.

Data will be analyzed using 2×2 contingency tables to compare AI screening results with echocardiography, and diagnostic performance metrics including sensitivity, specificity, positive predictive value, and negative predictive value will be calculated with 95% confidence intervals. Agreement between AI-assisted auscultation and echocardiography will be assessed using Cohen's kappa.

This study will provide evidence on the feasibility, accuracy, and scalability of AI-enabled smart stethoscopes for early SHD detection in school-based, low-resource settings.

Inclusion Criteria:

1. School students aged 10-18 years.
2. Able to cooperate with cardiac auscultation.
3. Student and parent/guardian provide written informed consent.

Exclusion Criteria:

1. Student or parent/guardian declines participation.
2. Inability or unwillingness to follow screening procedures or cooperate with cardiac auscultation.
3. Refusal to undergo standard transthoracic echocardiography or cardiology evaluation.
4. Previously diagnosed structural heart disease.
5. Chest wall deformities or skin conditions that may interfere with auscultation.
6. Fever ≥37.5 °C on the day of examination, or severe developmental delay or other conditions preventing cooperation with the examination.

Participants will be recruited from primary and secondary schools in Ruyang County, Henan Province, China. Screening will be conducted on-site at schools by trained personnel using AI-enabled smart stethoscopes. Students identified as having abnormal findings will be referred to the Cardiology Department of Ruyang County People's Hospital for confirmatory transthoracic echocardiography. A subset of students with normal screening results will also undergo echocardiography to estimate false-negative rates.