Abstract

Info

Author(s):: DENG Qishichao1; 2; WU Jiacheng1; 2; YU Jian1; 2; ZHANG Hua2; LI Chuantao3; LIN Dongmei4; CHEN Fuming2; 1. School of Medical Information Engineering, Gansu University of Chinese Medicine, Lanzhou 730000, China 2. Medical Support Center, 940th Hospital of Joint Logistics Support Force of Chinese Peoples Liberation Army, Lanzhou 730050, China 3. Naval Characteristic Medical Center, Naval Medical University (Second Military Medical University), Shanghai 200433, China 4. School of Electrical and Information Engineering, Lanzhou University of Technology, Lanzhou 730050, China

Keywords:: Keywords: heart sound signal deep learning classification algorithm neural network review

Abstract:: Abstract: Heart sounds are vital physiological signals that reflect cardiac status, and heart sound classification plays a crucial role in the early diagnosis of cardiovascular diseases. Compared with traditional methods that rely on handcrafted feature extraction, deep learning techniques can automatically learn representative features from heart sound signals, achieving superior classification accuracy and robustness. This study provides an overview of recent advances in deep learning-based heart sound classification by retrieving and analyzing relevant domestic and international research. This review systematical introduces the characteristics of heart sound signals, commonly-used datasets, and preprocessing methods, summarizes the application status of deep learning approaches, including convolutional neural networks, recurrent neural networks, Transformer-based models, and hybrid architectures, in heart sound classification, and further compares their model architectures, classification performance, and clinical applicability. Finally, the application prospects of deep learning in heart sound classification is discussed, aiming to promote the application of deep learning techniques in heart sound classification.