Publication: Automatic energy expenditure measurement for health science
No Thumbnail Available
Elsevier Ireland Ltd, Elsevier House, Brookvale Plaza, East Park Shannon, Co, Clare, 00000, Ireland
Background and objective: It is crucial to predict the human energy expenditure in any sports activity and health science application accurately to investigate the impact of the activity. However, measurement of the real energy expenditure is not a trivial task and involves complex steps. The objective of this work is to improve the performance of existing estimation models of energy expenditure by using machine learning algorithms and several data from different sensors and provide this estimation service in a cloud-based platform. Methods: In this study, we used input data such as breathe rate, and hearth rate from three sensors. Inputs are received from a web form and sent to the web service which applies a regression model on Azure cloud platform. During the experiments, we assessed several machine learning models based on regression methods. Results: Our experimental results showed that our novel model which applies Boosted Decision Tree Regression in conjunction with the median aggregation technique provides the best result among other five regression algorithms. Conclusions: This cloud-based energy expenditure system which uses a web service showed that cloud computing technology is a great opportunity to develop estimation systems and the new model which applies Boosted Decision Tree Regression with the median aggregation provides remarkable results. (C) 2018 Elsevier B.V. All rights reserved.
Human energy expenditure, Machine learning, Energy prediction, Physical-Activity, Neural-Network, Accelerometers, Prediction, Regression, Walking, Forest, Wrist, Hip