Heart Sound Classification

Heart sound classification uses machine learning to automatically diagnose cardiovascular diseases from audio recordings, aiming to improve diagnostic accuracy and efficiency compared to manual auscultation. Research focuses on developing robust classification models, often employing convolutional neural networks (CNNs), ensemble methods, and transformer-based architectures, with feature extraction techniques like Mel Frequency Cepstral Coefficients (MFCCs) and Time-Frequency Distributions (TFDs) playing a crucial role. Improved accuracy and the ability to handle noisy recordings are key goals, ultimately leading to potential advancements in early disease detection and improved patient care.