Advancements in Deep Learning for Cardiac Sound Classification: A Review on Architectures, Preprocessing Techniques and Datasets

The identification and classification of cardiac sound patterns play a crucial role in the timely detection and diagnosis of cardiovascular diseases. Advances in deep learning techniques have significantly contributed to this field, offering promising solutions for detailed and accurate analysis. This study provides an overview of advanced methodologies for classifying cardiac sounds using deep learning approaches. Various deep learning architectures, including convolutional neural networks, recurrent neural networks, and hybrid models, are discussed, highlighting their advantages and limitations in analyzing cardiac sounds. Additionally, the preprocessing techniques essential for preparing audio data for deep learning models are examined, along with a review of publicly available datasets commonly used in this area of research. The evaluation metrics for assessing model performance are also explored, and future directions for advancing this field are outlined. This study aims to guide researchers and practitioners in developing robust deep learning frameworks for monitoring cardiovascular health.