Search

• The input data is pre-processed using Morphological filtering. • Extended Empirical wavelet transformation for removing the baseline. • The pre-processed data are given to Ternary pattern and discrete wavelet transform. • The proposed BiGRU-AE-AD-ECG method is implemented in python. • It is examined under sensitivity, precision, recall, f-measure, specificity and accuracy. Heartbeat irregularity or arrhythmia of a patient with serious congestive cardiovascular breakdown is distinguished in this article, and the class not really settled utilizing long haul ECG information. Several existing methods on arrhythmia detection are utilized, which does not accurately classify the Arrhythmia and it takes higher computational time. Morphological filtering and Extended Empirical wavelet transformation preprocess the input data to remove noises and base line. Then these pre-processed data is fed to Ternary Pattern and Discrete Wavelet Transform (TP-DWT) related feature extracting to take relevant feature. Then, the extracted features are given to Bidirectional Gated Recurrent Unit with Auto Encoder (BiGRU-AE) to classify Arrhythmia into Normal, Atrial Fibrillation (AF), First-degree atrioventricular block (1-AVB), Left Bundle Branch (LBBB), Right Bundle Branch Block (RBBB), Premature Atrial Contraction (PAC), Premature Ventricular Contraction (PVC), ST-segment depression (STD) and ST-segment elevation (STE). The proposed BiGRU-AE-AD-ECG is activated in Python and its efficiency is measured under certain metrics, like sensitivity, precision, f-measure, specificity, accuracy, error rate, ROC and computational time. Here, CPSC 2018 dataset is used to validate the efficiency of the BiGRU-AE-AD-ECG method. The performance of BiGRU-AE-AD-ECG method provides 23.40 %, 18.78 % and 28.05 % higher accuracy, 20.41 %, 27.07 % and 18.80 % higher precision, 32.01 %, 27.40 % and 28.95 % higher sensitivity and 28.05 %, 23.41 % and 31.34 % higher specificity analyzed with existing models, like automatic cardiac arrhythmias categorization based on CNN and attention-based RNN network (CNN-RNN-AD-ECG), a multiple lead-branch fusion network under 12-Lead ECG for multiple class arrhythmia categorization (MLBFNet-AD-ECG), analysis of ECG-based arrhythmia identification scheme using machine learning (SVM-AD-ECG)respectively. [ABSTRACT FROM AUTHOR]