Advanced integration of 2DCNN-GRU model for accurate identification of shockable life-threatening cardiac arrhythmias: a deep learning approach

• DOI: 10.3389/fphys.2024.1429161
• Publication Date: 2024-07-12T04:59:57Z
• Authors: Abduljabbar S. Ba...; Shenghong Cao; Kaixuan Zhang; Samia Allaoua Che...; Rana Alnashwan; Mohammed Saleh Al...
• Keywords: Cardiac arrhythmia

Cardiovascular diseases remain one of the main threats to human health, significantly affecting quality and life expectancy. Effective prompt recognition these is crucial. This research aims develop an effective novel hybrid method for automatically detecting dangerous arrhythmias based on cardiac patients’ short electrocardiogram (ECG) fragments. study suggests using a continuous wavelet transform (CWT) convert ECG signals into images (scalograms) examining task categorizing 2-s segments four groups that are shockable, including ventricular flutter (C1), fibrillation (C2), tachycardia torsade de pointes (C3), high-rate (C4). We propose developing neural network with deep learning architecture classify arrhythmias. work utilizes actual data obtained from PhysioNet database, alongside artificially generated produced by Synthetic Minority Over-sampling Technique (SMOTE) approach, address issue imbalanced class distribution obtaining accuracy-trained model. Experimental results demonstrate proposed approach achieves high accuracy, sensitivity, specificity, precision, F1-score 97.75%, 99.25%, respectively, in classifying all shockable classes superior traditional methods. Our possesses significant clinical value real-life scenarios since it has potential enhance diagnosis treatment life-threatening individuals disease. Furthermore, our model also demonstrated adaptability generality two other datasets.