PATIENT-ADAPTED AND INTER-PATIENT ECG CLASSIFICATION USING NEURAL NETWORK AND GRADIENT BOOSTING

Abstract

Heart disease diagnosis is important non-invasive technique. Therefore, there exists effort to increase accuracy of arrhythmia classification based on ECG signals. In this work, we present a novel approach of heart arrhythmia detection. The model consists of two parts. First part extracts important features from raw ECG signal using Auto-Encoder Neural Network. Extracted features obtained by Auto-Encoder represent an input for the second part of the model, the Gradient Boosting and Feedforward Neural Network classifiers. For comparison purpose, we evaulated our approach with using MIT-BIH ECG database and also we fellow recommendations of the Association for the Advancement of Medical Instrumentation (AAMI) for ECG class labeling. We divided our experiment into two scenarios. First scenario represents classification task for patient-adapted paradigm and second one was dedicated to inter-patient paradigm. The measured results we compared with state-of-the-art methods and it shows that our method outperform state-of-the art methods in the Ventricular Ectopic (VEB) class for both paradigms and Supraventricular Ectopic (SVEB) class in inter-patient paradigm.