Electrocardiogram Heartbeat Classification Using Machine Learning and Ensemble Convolutional Neural Network-Bidirectional Long Short-Term Memory Technique

Automated classification of cardiac rhythms from electrocardiogram (ECG) signals is significant for diagnosis of cardiovascular dysfunctioning. A biggest challenge in automated ECG classification is to address the task's specific characteristics, such as time dependencies between observations and a strong class imbalance. To address these issues, this article proposes machine learning ensemble techniques (random forest, support vector machine, Xgboost, Adaboost, and stacked ensemble classifier) and an ensemble of convolutional neural network (CNN) and bidirectional long short-term memory (Bi-LSTM) architecture for classification of cardiac arrhythmias in ECG signals. The proposed model has been trained and tested on the MIT-BIH arrhythmias database, which contains a total of 109 443 ECG beats with 90 589 normal beats, 8039 supraventricular beats (SB), 7236 ventricular beats (VB), 2776 fusion beats, and 803 unknown beats, respectively. Here, in this article, we incorporate a synthetic minority oversampling technique for balancing dataset along with time-series feature extraction library for extracting features and an ensemble of CNN-LSTM to predict fast features of upcoming ECG signals to enhance the performance of heartbeat classification as compared to the state-of-the-art techniques. The achieved average accuracy (Acc), sensitivity (SE), precision (PE), and F1-score using stacked ensemble machine learning algorithm for detection arrhythmias heartbeat classes are 98.67%, 96%, 97.80%, and 96.80%, respectively. With the ensemble CNN-BiLSTM method, an average accuracy of 99.88% has been achieved. The SE and PE values for SB and VB beats are 84.40%, 99.79%, 96.40%, and 98.54%, respectively. Experimental results demonstrate that an ensemble CNN-BiLSTM technique outperforms as compared to the existing classification techniques in terms of sensitivity, accuracy, and precision.

I. Introduction

Cardiovascular diseases (CVDs) are the main causes of death around the world, of which 85% happened due to heart attack and affect the gross domestic product of the nation rigorously [1]. The electrocardiogram (ECG) is a noninvasive and cost-effective tool for the early detection of abnormal heart rhythm disorders in the cardiovascular system. A normal ECG wave consists of P-wave, QRS-wave, T-wave, and U-wave components. The U-wave has not been considered generally because its amplitude and heart rate depend on the T-wave amplitude. The P-wave represents atrial depolarization, the QRS complex represents ventricular depolarization, and the T-wave is associated with ventricular repolarization. The presence of arrhythmias in the heart is seen by changes that occur in the characteristics of these waves. An arrhythmia is an irregular occurrence of a heartbeat or pattern of heart rate rhythm. During the presence of abnormalities, the cardiac muscle cannot be able to pump the blood in the body accurately. Hence, insufficient blood flow can damage the heart, brain, and other organs. So, cardiologists examine the ECG recordings carefully to identify arrhythmias. Sometimes signals get unidentified due to small variations in the signal amplitude [2]. To analyze long ECG recordings, manual beat-to-beat exploration is time-consuming and inaccurate. The manual examinations of ECG recording also become error-prone and cumbersome for the cardiologists. Thus, there is an urgent need for automated ECG classification. The early recognition of cardiac arrhythmia is critical for effective investigation and treatment of CVDs.