Advancing Ischemia Detection: Multi-Model Classification of ST and T Changes in Ischemia ECG Signals

This study presents an innovative approach to is-chemia detection, focusing on the classification of ST and T changes within electrocardiographic (ECG) signals. Our methodology sequentially employs two distinct models to discern subtle electrocardiographic alterations indicative of ischemic events. We compared Logistic Regression, Multi-Layer Perceptron (MLP), Support Vector Machine (SVM), eXtreme Gradient Boosting (XGBoost), and Random Forest. Discrete Wavelet Transform (DWT) was used for data preprocessing. The models were trained and tested on the European ST-T Change Database, followed by evaluation on the QT Database. For ST change classification, SVM achieved exceptional performance with 100% accuracy on both the test set and the QT Database. However, for T change classification, XGBoost was surpassed by other algorithms. The best performing model for T change classification achieved 99.81% accuracy on the test set and 100% accuracy on the QT Database. These findings underscore the efficacy of our approach in enhancing ischemia detection, holding significant promise for improving clinical outcomes in cardiac care.