Automatic electrocardiograph diagnosis of myocardial ischemia with support vector machine

Abstract

ABSTRACT Background: Myocardial ischemia is a severe cardiac disease and it happens when the heart's blood flow is insufficient, which impairs its capacity to operate correctly and causes several systemic issues. A standard auxiliary for the clinical diagnosis of myocardial ischemia is an electrocardiograph (ECG). However, the typical static ECG cannot record the myocardial ischemia paroxysmal fragments quickly and accurately. The information contained in long-term ECG recordings is more abundant, but the large volume of data makes manual processing costly. Our aim is to propose an automated method for handling ECG signals and diagnosing myocardial ischemia. Methods: To get evidence of myocardial ischemia, dynamic ECG is frequently employed. This paper suggests a machine learning technique to create a classification model using support vector machine (SVM), automatically identifying the existence of myocardial ischemia fragments. This study comprises waveform identification, feature extraction, creation of data sets, model training, and classification. The Long Term ST Database, made available by PhysioNet, is used as the database for this research. Result: According to the final test results, the classification accuracy is 97.98%. The ECG signals can be automatically segmented, and the automated diagnosis of myocardial ischemia can achieve an accuracy rate of 97.98%. Conclusion: The results suggest that the method proposed in this paper can more precisely and practically identify myocardial ischemia pieces.