ABSTRACT
Heart arrhythmia is a condition in which the heartbeat of a person becomes irregular. It is rooted in a variety of causes, including changes in heart muscle, electrolyte imbalance, and so on. If this condition is not diagnosed in time, it can lead to fatal consequences, such as stroke, heart failure, and cardiac arrest. Electrocardiogram (ECG) is used to diagnose the heartbeat. An IoT (Internet of things)-based ECG acquisition system is used to acquire the real-time ECG signal from patients. The system records the ECG in real time and applies signal processing to get the dynamic features of the ECG from the RR intervals of the ECG signal. To find the R peak, the Pan–Tompkins algorithm is used. As per this research study, various deep learning techniques like neural networks (NNs)/multi-layer perceptron (MLP), convolutional neural network (CNN), and long short term memory (LSTM) and IoT are implemented and compared in an effort to classify one subset of MIT-BIH dataset, which consists of 100,000 samples of two-channel ambulatory ECG scans. The implemented CNN model achieved a 92% macro average, while the MLP model got an 86% macro average. The weighted average was lowered due to fewer number of samples for certain classes.
