Paper Title
DEEP LEARNING MODEL COMPARISON FOR ECG-BASED CARDIOVASCULAR DISEASE DIAGNOSIS

Abstract
This work proposes a non-invasive ECG-based approach for early detection of heart disorders. By converting ECG signals into twelve-lead images, the method enables deep learning classification using SqueezeNet, AlexNet, and a custom CNN into four categories: arrhythmias and myocardial infarction risk/history. The research addresses the global prevalence of cardiovascular diseases by leveraging a non-invasive, needle-free ECG modality for early cardiac disorder detection. ECG signals are transformed into twelve-lead image representations, facilitating deep learning analysis through SqueezeNet, AlexNet, and a custom convolutional neural network. The models achieve classification across four categories, encompassing diverse arrhythmias and indicators of myocardial infarction risk or history. Keywords - Cardiovascular Diseases, Heart Conditions, Mortality, Timely Prediction, Electrocardiogram (ECG), Deep Learning, Transfer Learning, Neural Networks, Squeeze Net, Alex Net.