Automated Phonocardiography Analysis in Adults

Background: Computer aided auscultation in the differentiation of pathologic (AHA class I) from no- or innocent murmurs (AHA class III) via artificial intelligence algorithms could be a useful tool to assist healthcare providers in identifying pathological heart murmurs and may avoid unnecessary referrals to medical specialists. Objective: Assess the quality of the artificial intelligence (AI) algorithm that autonomously detects and classifies heart murmurs as either pathologic (AHA class I) or as no- or innocent (AHA class III). Hypothesis: The algorithm used in this study is able to analyze and identify pathologic heart murmurs (AHA class I) in an adult population with valve defects with a similar sensitivity compared to medical specialist. Methods: Each patient is auscultated and diagnosed independently by a medical specialist by means of standard auscultation. Auscultation findings are verified via gold-standard echocardiogram diagnosis. For each patient, a phonocardiogram (PCG) - a digital recording of the heart sounds - is acquired. The recordings are later analyzed using the AI algorithm. The algorithm results are compared to the findings of the medical professionals as well as to the echocardiogram findings.