Disease Detection Using Heart Sound Classification

Heart sound classification plays a critical role in the early detection and diagnosis of cardiovascular diseases. This project aims to develop a disease detection system by analyzing heart sounds to classify them into three categories: artifact, murmur, and normal. Utilizing state-of-the-art deep learning algorithms such as Convolutional Neural Networks (CNN), Long Short-Term Memory networks (LSTM), and ResNet, this system processes audio heart sound data to detect potential anomalies. The dataset for this project is sourced from Kaggle, which contains labeled audio files for heart sound signals. By training the models on this dataset, the system is capable of distinguishing between healthy heartbeats and those with murmurs or artifacts, providing valuable support for clinicians in making accurate and timely diagnoses. The system is designed to accept audio file uploads, classify the heart sound, and provide a reliable diagnosis based on the classification. The outcome of this research is a scalable, efficient, and automated tool that enhances early detection of heart diseases, offering significant potential for improving patient care and outcomes. Keywords: Heart sound classification, CNN, LSTM, ResNet, artifact, murmur, normal, disease detection, deep learning, audio analysis, cardiovascular diseases.