Automatic Classification of Cardiac Disorders Using MLP Algorithm

World Health Organization (WHO) 2018 report says that 31% of global deaths are due to cardiovascular diseases. If no initiative is taken to check this most predictable and preventable among all chronic diseases, the world will have to suffer from billions of heart patients. India has the highest incidence of heart related diseases in the world and also it has one doctor for every 10,189 people though WHO suggests one doctor for 600 people [11]. The main aim of this work is to enable the doctors for early detection of cardiac disorder and save the life of heart patients. This work may be suitable for Tele-Cardiology application of Tele-Medicine which has a unique capacity to improve the health care service to millions of rural people. Cardiac arrhythmias classification has become an active research area over the last two decades. Further, from a literature survey, it is identified that finding an efficient feature scheme is a major challenge to enhance classification accuracy of cardiac disorders. Existing feature extraction methods may not be sufficient to detect the possibility of cardiac disorders as ECG is a quasi-periodic, nonlinear and non-stationary signal. Hence, in this paper, it is proposed to use a wavelet transform technique to extract ECG features. These features are used to classify seven types of cardiac arrhythmias (Cardiac Ischemia, Ventricular Tachycardia, Supra Ventricular Tachycardia, Ventricular Fibrillation, Atrial Fibrillation, Sudden Cardiac Arrest including Normal Sinus Rhythm). Multi-Layer Perceptron neural network classifier is proposed for this classification. Neural networks are strongly capable of learning and prediction which makes them an efficient tool to deal with nonlinear problems depending on their training data set size. With this proposed work, the classification accuracy is obtained as 98.8% and the medical statistics are obtained as Sensitivity (S) 97.6%, Positive Predictivity (Pp)-98.7% and Specificity (Sp)-99.7%. Further, with this efficient wavelet feature scheme, a Prototype Cardiac Alert System has been implemented using Arduino Uno and GSM SIM 900A for remote monitoring of the heart patients. A standard MIT-BIH ECG database, Waikato Environment for Knowledge Analysis (WEKA) and Matlab R2013a software have been used for the implementation.