Detection of Cardio Vascular abnormalities using gradient descent optimization and CNN

Purpose: The purpose of this study is to propose an advanced methodology for automated diagnosis and classification of heart conditions using electrocardiography (ECG) in order to address the rising death rate from cardiovascular disease (CVD). Methods: Buffered ECG pulses from the MIT-BIH Arrhythmia dataset are integrated using a multi-modal fusion framework, refined using Gradient Descent optimization, and classified using the K-Means technique based on pulse magnitudes. Convolutional Neural Networks (CNNs) are used to detect anomalies. Results: The study achieves an average accuracy of 98%, outperforming current state-of-the-art methods. Sensitivity, specificity, and other metrics show significant improvements. The results also show the type of Cardiovascular disease detected using Confusion matrix plots. Conclusion: The proposed methodology demonstrates the utility of advanced machine learning, particularly deep learning, in the assessment of cardiovascular health. Based on the MIT-BIH Arrhythmia dataset, this study contributes to the development of accurate and efficient diagnostic tools for addressing urgent cardiac health challenges.