Cardiovascular Disease Prediction and Classification using Modified Neural Network and Support Vector Machine

Abstract

Arrhythmia beat classification is most often used to detect electrocardiogram (ECG) abnormality to identify the heart problems. Arrhythmia seeming with the occurrence of abnormal activity is efficiently recognized and classified with diverse categories. In the proposed method, Median filter is applied for pre-processing to get signal smoothing; morphological features are extracted from P-QRS-T waves, and separated from the selected ECG segment by using Discrete Wavelet Transform. The   features are applied to Probabilistic Neural Network(PNN) and automatic diagnosis results are evaluated. The authenticated arrhythmia datasets are taken from MIT-BIH source page, in which ECG signals are segregated into the seven recommended classes: Normal beats (N), Supraventricular beats (SVEBs), Ventricular ectopic beats (VEBs), Premature Contraction, Bradycardia, Tachycardia, and Atrial flutter. The proposed technique predicts the classes with accuracy of 94.7%, sensitivity of 95% and specificity of 94.4%.