Continuous Heartbeat Prediction Using a Face Recognition Algorithm Ahmed A. Alsheikhy 1* , Yahia F. Said 1 , Tawfeeq Shawly 2 1 Electrical Engineering Department, College of Engineering, Northern Border University, Arar 91431, Saudi Arabia 2 Electrical Engineering Department, Faculty of Engineering at Rabigh, King Abdulaziz University, Jeddah 21589, Saudi Arabia Corresponding Author Email: aalsheikhy@nbu.edu.sa https://doi.org/10.18280/ts.390506 ABSTRACT Received: 14 April 2022 Accepted: 20 September 2022 Health providers use the ECG machine to get information about the heart. This information plays a significant role since it tells them about the status of the heart. The ECG machine presents this information in a waveform. During the Covid-19 pandemic, all governments have placed numerous rules and policies to protect people from the virus and from spreading it. One of the rules and policies is to prevent touching surfaces in public places. However, in health care centers, touching surfaces can’t be avoided completely since there is a need to touch them or place some wires on the human body such as placing wires to use the ECG machine. In Saudi Arabia, the government has placed a policy in all its buildings, public places, and the private sector to measure the temperature at the entrance. Due to this situation, the idea has come into mind to have a touchless method to measure the heartbeat rate. In this paper, proposing a feasible and reliable method to estimate a continuous heartbeat rate is presented. It uses a face recognition approach to predict the heart pulse continuously in real-time according to colors intensity measurement. Using a segmentation algorithm is involved since the approach takes its input from a video or an image. Several experiments have been conducted on volunteers to verify the obtained results and measure their relative errors. Consequently, the errors are less than 7% which is quite acceptable. At the end of this article, a comparative assessment is performed between the presented approach and some works from literature. This assessment is conducted based on the methodologies being utilized and applied and Mean Absolute Error (MAE). Furthermore, it shows whether those methods require physical contact or not. The obtained results indicate that the implemented system herein outperforms other state-of-the-art methods. Keywords: heartbeat, artificial intelligence, face recognition, heart rate, machine learning, image segmentation, cardiology, cardiovascular 1. INTRODUCTION Measuring a human vital sign such as heart pulse is a very crucial task. Physicians and healthcare providers rely on it to obtain some information about patients’ conditions. High heartbeat rate or low rate indicates that there is an issue and physicians should investigate to see the real causes. In most cases, physicians and healthcare providers use equipment to measure the heartbeat rate. ECG refers to Electrocardiogram which is an electrical waveform to display signals that are provided by the heart [1, 2]. These signals represent the heart’s activities. The wave is produced by placing adhesive electrodes on the skin. The purpose of these electrodes is to predict small electrical changes that occur due to cardiac muscle movements [1-4]. Those movements are the result of repolarization and depolarization during every cardiac cycle [3-5]. ECG refers to the heart function which is represented by the relationship between amplitude and phase as depicted in Figure 1. Any abnormal activities or functions of the heart are seen as a deviation from the standard amplitude-phase relationship [2]. Consequently, this deviation indicates the presence of any abnormal activity [2, 6-9]. Normal ECG signal consists of the following components as depicted in Figure 1, P wave, QRS complex, T wave, R wave and U wave. The P wave refers to the atrial depolarization [2, 4, 5]. The ventricular depolarization is represented by the QRS complex. The T wave describes the ventricular repolarization [2, 4]. The R wave refers to the peak of the signal and represents the positive deflection [2, 6, 10]. And the U wave refers to the papillose muscle repolarization. The Q and S waves as shown in Figure 1 occur before and after the R wave respectively. In Figure 1, PR interval refers to the time needed for the electrical impulse to go from the sinus node to the AV node [2] and it is represented by the orange color. The QT interval, which is depicted by the blue color in Figure 1, represents the ventricular depolarization, repolarization and contraction [8, 9-12]. Recently, researchers have developed and implemented contactless systems to monitor heartbeat rate continuously. These systems and applications depend on contactless sensors to measure the heart rate [3, 4]. The contactless applications provide no restrictions on mobility. Thus, users are free to move anywhere without being afraid of losing the measurement. In addition, these systems and applications are usable for patients who suffer from skin irritations [4, 13]. The contactless systems provide real-time data for the heart rate and are required for the timely detection of abnormal activities [3-5]. Healthcare providers and patients rely on these systems to detect any abnormal activity which could lead to dangerous conditions such as drowsiness [4, 13-17]. Traitement du Signal Vol. 39, No. 5, October, 2022, pp. 1501-1506 Journal homepage: http://iieta.org/journals/ts 1501