Deep Learning Approach on Multimodal Data for Myocardial Infarction Prediction

This research introduces a novel machine learning methodology for classifying electrocardiogram (ECG) images, integrating deep learning models like VGG16, Inception V3, and a custom CNN, alongside ensemble methods such as XGBoost and AdaBoost with RandomForest. The primary objective is to enhance the accuracy of automated ECG analysis, crucial for early detection of cardiovascular diseases (CVDs). By employing comprehensive data preprocessing and augmentation techniques, the models are trained on a diverse dataset encompassing various cardiac rhythm categories. Preliminary results indicate superior performance of the custom CNN model, underscoring the potential of machine learning in improving cardiac diagnostic processes and patient outcomes. Furthermore, patient analysis using a comprehensive dataset containing clinical attributes relevant to heart health supplements the research, aiming to identify key risk factors associated with heart disease and explore their predictive value in diagnosing CVDs. This integrated approach highlights the transformative potential of machine learning in cardiovascular healthcare, promising early detection and personalized treatment strategies for improved patient care.