DEVELOPMENT OF A MACHINE LEARNING MODEL FOR AUTOMATED DIAGNOSIS OF ECG ABNORMALITIES

Abstract

Early diagnosis of Cardiovascular diseases (CVD) remains essential because these diseases represent the main global reason for mortality and disease incidents. Traditional ECG analysis methods struggle to identify minor and intricate abnormalities within extensive databases because Electrocardiograms (ECGs) serve as standard tools to inspect heart health conditions. The research creates a hybrid machine learning model for automated ECG abnormality detection which integrates deep learning features of CNN (1D Convolutional Neural Network) for feature extraction with classifier chain ensemble modeling for multi-label classification. The experimental results based on the PTB-XL database demonstrated an overall accuracy rate of 87.79% together with a weighted accuracy of 95.98% through model training and evaluation. This performance exceeded standard methods. The main contribution of this study includes connecting a machine learning algorithm to a web-based interface which enables healthcare professionals especially those in limited resource environments to submit single-lead ECG images that produce instant diagnostic results. This integration facilitates faster, more efficient ECG analysis with a response time of approximately 22 seconds per image. The hybrid system represents progress in medical diagnostic workstreams and expands automatic cardiovascular disease detection through its monitoring platform despite facing difficulties in detecting uncommon abnormalities. Studies prove that machine learning-based ECG diagnostic systems can successfully enter clinical settings where they create a practical and quick method to enhance cardiovascular treatment.