Deep Learning for Heart Sound Analysis: A Literature Review

Heart sound auscultation is a physical examination routinely used in clinical practice to identify potential cardiac abnormalities. However, accurate interpretation of heart sounds requires specialized training and experience, thereby limiting its generalizability. Deep learning, a subset of machine learning, involves training artificial neural networks to learn from large datasets and perform complex tasks related to intricate patterns, such as disease diagnosis, event prediction, and clinical decision-making. Over the past decade, deep learning has been successfully applied to heart sound analysis with remarkable achievements. Meanwhile, as heart sound analysis is gaining attention, many public and private heart sound datasets have been established for model training. The massive accumulation of heart sound data improves the performance of deep learning-based heart sound models and extends their clinical application scenarios. In this review, we will compile the commonly used datasets in heart sound analysis, introduce the fundamentals and state-of-the-art techniques in heart sound analysis and deep learning, and summarize the current applications of deep learning for heart sound analysis and their limitations for future improvement. ### Competing Interest Statement The authors have declared no competing interest. ### Funding Statement This study was funded by the National Natural Science Foundation of China (No. 62102008) and Peking University People's Hospital Scientific Research Development Funds (RDJP2022-39). ### Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes The data availability is based on each literature discussed in the review. * DL : deep learning PCG : phonocardiogram AS : aortic stenosis MR : mitral regurgitation MS : mitral stenosis MVP : mitral valve prolapse STFT : Short-Time Fourier-Transform MFCCs : Mel-frequency Cepstral Coefficients CEEMD : Complementary Ensemble Empirical Mode Decomposition TQWT : Tunable-Q Wavelet Transform CWT : Continuous Wavelet Transform CNNs : convolutional neural networks RNNs : recurrent neural networks NN : neural network ML : machine learning VHD : Valvular heart disease DNN : Deep Neural Network CHD : Congenital heart disease ASD : atrial septal defects VSD : ventricular septal defects PDA : patent ductus arteriosus TGNN : Time Growing Neural Network LVEF : left ventricular ejection fraction GRU : gated recurrent unit HFpEF : Heart failure with preserved ejection fraction HFrEF : Heart failure with reduced ejection fraction LVDD : left ventricular diastolic dysfunction CAD : Coronary artery disease RHD : Rheumatic heart disease BP : blood pressure PH : pulmonary hypertension PAP : pulmonary artery pressure