Certain investigation on remote Detection of arrhythmias using Meta heuristics optimization Techniques

Abstract

An Internet of Things (IoT) based healthcare application is now newlineprevalent as preventive care and can further offer various benefits using newlineconnected devices such as monitoring patients symptoms and conditions. newlineThe non-stationary Electrocardiogram (ECG) signals have been widely newlineutilized for heartbeat assessment to diagnose cardiovascular diseases. A newlinenetwork with numerous small sensors and self-organization ability for these newlinesensors is defined as the Wireless Sensor Network (WSN). For the prediction newlineof cardiovascular diseases, a WSN platform that employs a WSN-enabled newlineECG telemetry system will involve the following steps: the ECG signal s newlineacquisition, the ECG signal s processing, and alerting the physician in case of newlineany emergencies. This system will aid the physician in the early and accurate newlineanalysis of heart diseases. In this work, an IoT -enabled ECG monitoring newlinesystem is being developed where WSN is used to acquire the ECG signals and newlineprocess the ECG signal to identify cardiovascular disease. To improve the newlineframework s performance, the routing of the WSN is optimized. Similarly, to newlineenhance the classification of ECG, the classifier is optimized. newlineThis work has employed the Artificial Bee Colony (ABC) and the newlineGrey Wolf Optimizer (GWO) for optimizing the clustering to boost the WSN newlinerouting and network longevity. This work presents a Convolutional Neural newlineNetwork (CNN) technique for automatically detecting the distinct ECG newlinesegments. The features are fed as inputs in Feed Forward Neural Networks newline(FFNN) and Recurrent Neural Networks (RNN) to classify the ECG as newlinenormal or arrhythmia. FFNN, or Multi-Layer Perceptron (MLP) neural newlineiv networks, are the most renowned neural networks in which there is the newlineconstruction of input-output relations through adjustment of the network s newlineconnection weights. The RNNs are particularly feasible with sequential data newlinesince each one of its neurons is able to employ its internal memory to store newlineinformation related to the preceding input newline newline