| dc.description.abstract | Cardiovascular disease is the leading cause of death globally, with a death toll reaching approximately 17.9 million people annually according to the World Health Organization (WHO). Early detection of heart disorders, including arrhythmia, is key to providing faster and more appropriate treatment. This study aims to prove that early detection of heart disease is possible through the development of a portable, Internet of Things (IoT)-based Electrocardiogram (ECG) device capable of wireless and real-time heart rate monitoring and classification. Data from the MIT-BIH Arrhythmia Database was used as training and validation data using the hold-out method, while testing was conducted with local data collected from university students as a representation of users in everyday life. The classification system was developed in two models: binary classification (Normal and Abnormal) and multiclass classification (Normal, Abnormal, Potential for Arrhythmia, and Highly Potential for Arrhythmia) by considering three physical activity conditions: sitting, standing, and running. The artificial intelligence models used, Random Forest and XGBoost, achieved accuracy of up to 99.8% for binary classification and 98.03% for multiclass classification with an execution time of 0.55 seconds. These results demonstrate that early detection of potential heart problems can be implemented effectively in everyday life through effective and efficient portable devices. | en_US |
