A gradient-boosting-regression based physical health evaluation model for running monitoring by using a wearable smartband system

Abstract

Due to insufficiency in data collection and analysis by using the existing wearable devices in physical fitness tests for adolescents, this paper presents a machine learning based physical health evaluation model for running activity monitoring, in which a gradient boosting regression algorithm is employed to process physiological data collected from a set of smartbands developed by ourselves. First, we collect two kinds of dynamic data including heart rate and acceleration when students wear our smartbands in a normal running test. Next, several key features closely related to the physical health status are extracted from the dynamic data. A gradient boosting regression (GBR) algorithm is then utilized to train a physical health evaluation model and calculate out a comprehensive score representing physical health status of each student for reference. Experiment results show that not only does the proposed model with GBR achieve higher evaluation accuracy than the one with another typical algorithm—support vector regression (SVR), but it also provides a promising solution for future physical health evaluation by using a machine-learning-model based intelligent computing instead of traditional empirical-model based manual calculation.