Designing a low-cost real-time group heart rate monitoring system

The aim of this paper is to present two different architectures for evaluating individual or group stress levels. First, the cloud architecture is presented in which a client device is used simply to detect an ECG (electrocardiographic) or PPG (photoplethysmographic) signal and transmit it in raw form to a server, where all the signal processing and evaluation is then carried out. The second architecture consists of a stand-alone client device that measures the signal and also performs signal processing, indicating the level of stress. Here the server merely receives the processed data and is mainly used to perform additional data analysis while providing a platform to show the stress (and other) evaluation results. We compared both architectures regarding the amount of data transmitted between the client and the server, the power and memory consumption of the client, and accuracy of the QRS algorithm detection where QRS refers to the heart's ventricular depolarisation and is represented by three deflections observed on a typical electrocardiogram. The QRS detection algorithms were tested with the MIT-BIH Arrhythmia Database and real-life data. The results indicate the accuracy of the QRS detection on the stand-alone device compared to that on the server side. A hybrid approach is more user-friendly because the stress-level results are displayed directly by the device rather than by an additional device such as a smart phone, whereas the server side allows for the long-term analysis of stress levels. These two client devices also have comparable prices, although the price of the stand-alone device is somewhat higher due to the graphical stress-level indicators. Therefore, we propose the hybrid approach when one is seeking a high-quality, miniature and low-cost solution that directly shows the stress level in real-time.