Deep Learning Techniques for Multi-Class Classification of Cardiac Sounds

This research implements deep learning techniques to classify cardiac sounds based on mel-frequency spectrogram features, aiming to enhance diagnostic accuracy for heart-related abnormalities. Utilizing a diverse dataset, the study highlights the Bidirectional Long Short-Term Memory (BiLSTM) model's exceptional performance, emphasizing the role of bidirectional information flow in capturing sequential patterns. The findings underscore profound learning potential, especially Long Short- Term Memory (LSTM) models, for improved cardiovascular health diagnostics. The research highlights specific performance metrics for each implemented model, presenting a comprehensive comparison of other deep learning models.