<![CDATA[Cardiac arrhythmia is a leading cause of sudden death, requiring rapid detection in emergency department nursing care. Manual interpretation of electrocardiograms (ECGs) is often time-consuming and prone to human error. This study aims to develop a lightweight and accurate automated arrhythmia classification system using a deep learning approach. The proposed method transforms one-dimensional ECG signals into two-dimensional visual representations in the form of spectrograms using the Short-Time Fourier Transform (STFT). The 48 x 48 pixel resolution spectrogram images are then classified using a 4-block Convolutional Neural Network (CNN) architecture to detect four rhythm classes: Normal, Atrial Fibrillation (AFib), Ventricular Tachycardia (VTach), and Supraventricular Tachycardia (SVT). Test results show that the model achieves 100% accuracy, precision, recall, and F1-score on the test data. The novelty of this research lies in the integration of Confidence Score Thresholds (>90%, 70-90%, <70%) as a clinical decision support system for nurses. With minimal computational overhead, this system can operate in real time on standard devices, supporting accelerated medical response without requiring extensive hardware infrastructure.]]>
Copyrights © 2026
