Arrhythmia Classification: A Review of Machine Learning-based Techniques for Electrocardiogram Signals using Wearable Technologies for the Elderly

The electrocardiogram (ECG) is one of the most important medical exams that could display, in form of a trace on a paper, the electrical activity of the heart. Different wearable technologies have been lately developed to monitor and transmit the ECG or other medical parameters of a patient to the medical professionals and are integrated into smartwatches or bracelets in order to be easily used by all types of patients, including the elderly. As they are more exposed to cardiovascular diseases, machine learning»based methods have been implemented, throughout the time, to better and faster classify ECG signals into normal or an'hythmia-related for a quick and correct response. Considering the fact that atrial fibrillation is the most known arrhythmia that could increase the risk of strokes and heart failures, an early detection and prevention can be vital, especially for an elder patient. Therefore, the aim of this article is to present some of the most recent methods of collecting the ECG signals using wearable devices and different machine learning-based techniques used for arrhythmia classification. Moreover, there are highlighted some methods meant to be used for the ECG data acquisition in the ongoing project, entitled “RO-SmartAgeing”, which aims to monitor and evaluate the daily activities and medical parameters of an elder patient.