Graphene Nanoribbon Electrode for real time ECG signal analysis by machine learning methods

Abstract

The real time Electro Cardio-Gram (ECG) signal is acquired using newly designed Graphene Nano Ribbon (GNR) dry electrode. The GNR system is designed using Arduino Microcontroller and AD8232 sensor to acquire real time ECG signal. GNR electrode is designed by drop casting method, which has the good sensitivity and low skin to electrode impedance to acquire the real time bio-potential signal. Ag/AgCl electrode requires gel which dries out and leads to signal degradation. GNR electrode used to acquire ECG signal, which does not lead to skin irritation and degradation of signal as its dry electrode. Real time ECG signals are acquired by 20 patients of different age groups are considered for analysis. Adaptive Least Mean Square (LMS) algorithm used to remove the power line interference and motion artifact of the real time ECG signal. The Scaled Rolling mean algorithm used for R-peak detection. The ECG signal parameters like R peak and R-R interval are considered for classification. The Machine learning algorithms like Support Vector Machine (SVM), K-Nearest Neighbor (KNN) and Naïve Bayes algorithm are compared. ECG signals signal to noise ratio(SNR) acquired by GNR electrode is 59.99 dB and sensitivity of 0.98. The accuracy obtained by different algorithms are KNN is 96%, SVM is 93% and Naïve Baye's algorithm is 90%. Hence GNR electrode can be used to acquire real time ECG signal for classiffication. The KNN method has highest accuracy to classify the acquired real time ECG signal by GNR electrode.