<![CDATA[An electrocardiogram (ECG) can detect heart abnormalities through signals from the rhythm of the human heartbeat. One of them is arrhythmia disease, which is caused by an improper heartbeat and causes failure of blood pumping. In reading ECG signals, a common problem encountered is the uncertainty of the prediction results. An accurate and efficient heart defect classification system is needed to help patients and healthcare providers carry out appropriate therapy or treatment. Several studies have developed algorithms that are more effective in Machine Learning (ML) in automatically providing initial screening of patients' heart conditions. This study proposed the Long Short-Term Memory (LSTM) method as a classifier of heart conditions experienced by humans and Continuous Wavelet Transform (CWT) as a feature extractor to eliminate noise during data collection. CWT and LSTM methods are believed to perform well in feature extraction and classification of ECG signals. The dataset used in this study was taken from the MIT-BIH Arrhythmia Database. The results of this study successfully extracted ECG signals using CWT, thus improving the understanding of ECG characteristics. This research also succeeded in classifying ECG signals using the LSTM method, which obtained an accuracy training value of 98.4% and an accuracy testing value of 94.42 %.]]>
