A Real-Time Cardiac Arrhythmia Classification Using Hybrid Combination of Delta Modulation, 1D-CNN and Blended LSTM.

The Real-time wearable Electrocardiogram (ECG) monitoring device is a perfect choice for assisting in detecting cardiovascular disease. A novel ECG beat classification algorithm is presented for continuous heart monitoring on low-processing-capacity wearable devices. We discuss the different wearable wristwatch monitoring system, which allows for continuous 24-h heart rate monitoring. This paper introduces a novel method for classifying arrhythmias based on deep learning. The method relies on QRS/PT detection, Sigma-Delta Modulation (SDM), One-Dimensional Convolution Neural Networks (1D-CNN) algorithms. The QRS/PT wave detection system is based on the 1D-CNN and SDM framework with local minimum and local maximum point algorithms. We proposed a Long Short-Term Memory (LSTM) recurrent neural network with a blend classifier. The classifier combines two small LSTM networks' predictions using two different features directly extracted from 1D-CNN and SDM bitstreams. The proposed model is evaluated by detecting QRS/PT waves and classifying arrhythmias using the MIT-BIH Arrhythmia Database. Five different classifications are performed and evaluated by the AAMI standard: N, F, Q, S, and V. The values for accuracy, positive predictivity, sensitivity, and F1-score are 99.56%, 96.5%, 93.87%, and 95.18%, respectively. The proposed algorithm detects QRS/PT in approximately 2050 ms and classifies each heartbeat in approximately 40–60 ms with wearable devices, and it consumes 1.5 μ w total power. The results indicate that our proposed system outperforms previous research in accuracy and meets both computation time and low power consumption requirements. [ABSTRACT FROM AUTHOR]