ECG-based Heart Arrhythmia Recognition using Enhanced Extreme Learning Machine

Abstract

Reliable classification of electrocardiogram (ECG) signal patterns is fundamental for enabling early diagnosis and effective clinical evaluation of cardiovascular abnormalities. In this paper, an innovative hybrid system is proposed for accurate classification of cardiac arrhythmias using ECG signals as input features. We present a deep learning-based method which utilizes the CNN model for complex features extraction, then integrates an Enhanced Extreme Learning Machine with Leave-One-Out (LOO) cross validation classifier to obtain both correct and stable models during training. In tests on the MIT–BIH Arrhythmia Database, the accuracy of CNN-EELM-LOO model was 99.81% and it exhibited improvement over classic machine learning and nowadays deep learning processes. The Leave-One-Out cross-validation technique integrated into decision model refinements provides an additional method to increase overall robustness of the model while reducing potential for random initialization effects or bias that can occur even with repeated training. This research demonstrates that the hybrid EELM-LOO architecture functions as an effective tool for automated arrhythmia detection, with strong potential for integration into clinical decision support systems.