Cardiac Activity Classification using an E-Health App for a Wearable Device

In the paper, authors discuss the development and performance evaluation of an e-health Android App, which allows to monitor and classify ECG-related features. The mobile platform is composed of a wearable and non-obtrusive sensor and a smartphone, which collects and process data during the daily life. The main goal of this work has been the development of a real-time, wireless monitoring tool that is not related to an indoor environment. Five physiological quantities are continuously measured: Heart Rate (HR) and 4 ECG time intervals (i.e., QT, ST, PR, QR); together with the complete ECG waveform, these parameters are displayed in real-time, while the user is performing different activities, indoor or outside. A visual feedback (i.e., colour bars, according to the level of the measured quantities) is provided to the user, together with the possibility to save data for further analysis. A pilot study has been conducted to evaluate the accuracy and the uncertainty of the data computed in real-time by the mobile application, to those derived from a reference system (i.e., ECG signal coming from a standard electrocardiograph), in post-processing. Results have shown a good agreement with gold standard instrumentation. The Bland-Altman test has identified a high correlation between the data computed by the App with respect to the ones from the gold standard (i.e., R
²
equal or higher than 90%), with a slight underestimation of the time intervals (3.6±1.5 ms for QT, 4.1±1.7 ms for ST, 3.0±1.2 ms for QR and 11.9±1.9 ms for PR interval).