Abstract: In the era of smart life, wearable devices are revolutionalizing health care. Estimating the heart rate of a person to monitor their health and fitness is an integral part of using wearable devices that contain Photoplethysmography (PPG) sensors. Though the PPG sensor gives an easier estimation of heart rate when compared to electrocardiography (ECG), the motion artefacts can act as an impediment affecting the accuracy and thus 3D accelerometer sensor is also used to combat this. This paper proposes a novel approach in order to classify and estimate heart rate while performing activities using machine learning models. The results of which show that an accuracy of 96.67% is obtained when the Random Forest classifier is used followed by other ensemble classifiers like Bagging and AdaBoost classifiers.
Keywords: Wearable devices, PPG sensor, 3D Accelerometer, heart rate estimation, activities.
| DOI: 10.17148/IJARCCE.2021.106104