Preliminary Study on Real-Time Phonocardiogram Signal Acquisition and Analysis Using Machine Learning and IoMT for Digital Stethoscope

Cardiovascular diseases currently pose the greatest threat to human health and future predicament is uncertain. Since most heart-related problems are reflected by the small variations in the heart’s sounds, quality research into heart rhythm will provide valuable health information that may identify and treat cardiac-related issues and disorders. Driven by the critical need for early and precise cardiac diagnostics, this study tackles the challenges of detecting subtle variations in heart sounds, such as S3 and S4, which are essential for identifying complex cardiovascular diseases (CVDs). This study introduces a real-time integrated system for heart sound detection, acquisition, analysis, and remote communication. A digital stethoscope was developed to facilitate computer-assisted auscultation, capturing heart sounds from the skin’s surface. These sounds are filtered, amplified, and transmitted to healthcare providers through the Internet of Medical Things (IoMT), enabling remote visualization and auscultation. Advanced signal segmentation and classification were performed using machine learning algorithms, including K-Nearest Neighbor (KNN) and Support Vector Machine (SVM). The results demonstrate that SVM and Naive Bayes Classifier (NBC) achieve high classification accuracies of up to 92.5% for normal heart sounds and 98.63% for abnormal sounds. This system enhances diagnostic precision, supports remote cardiac monitoring, and reduces the need for highly specialized personnel during preliminary assessments. By leveraging technological advancements, this research establishes a foundation for scalable and cost-effective solutions in cardiac care.