Risk Factor Analysis for Early Prediction of SCD in ECG Using BI-ANFIS

Abstract

One of the common outcomes of coronary artery disease (CAD) is Sudden Cardiac Death (SCD). Sudden cardiac arrest can lead to death without immediate treatment. The risk factors should be examined to analyse the risk level of SCD earlier. Nevertheless, it is challenging to analyse such parameters in Electrocardiogram (ECG) signals. Hence, the risk factors are analyzed in this article to predict the risk level of SCD utilizing a Bilinear Interpolation-based Adaptive Neuro-Fuzzy Inference System (BI-ANFIS) in ECG signal. Primarily, the ECG signal is taken as input and pre-processed for noise removal and frequency modulation correction. Afterward, by utilizing the Pan-Tompkins-based Hidden Markov Model (PT-HMM), the signal intervals are segmented. Thereafter, for analysing the first risk factor, the features from segmented waves are extracted, selected, and validated; also, CAD is predicted utilizing Soft plus error function-based Multi-Layer Perceptron (Serf-MLP). Concurrently, the smoke and QTc risk factors are evaluated. Now, the fourth risk factor is analyzed by the extraction of ST wave, J-wave selection, and type identification. Lastly, the BI-ANFIS is utilized to predict the SCD level of the ECG wave features grounded on the analyzed factors. Hence, the proposed technique’s superiority over the conventional approaches is proved by the final outcomes.