Deep Learning for Energy Expenditure Prediction in Pre-School Children

Abstract

Accurate monitoring of physical activity and its respective energy expenditure can help to reduce sedentary behaviour in preschool children. This paper 1) proposes the use of deep learning to effectively predict energy expenditure from body-worn accelerometers in pre-school-aged children, and 2) compares the deep learning approach to conventional supervised machine learning and simplified regression approaches. Eight preschool-aged children (5.23 ± 0.75 years old) performed 10 simulated free-living activities ranging in intensity from sedentary to vigorous. Participants wore accelerometers on both wrists and right hip, along with a portable metabolic system for direct assessment of energy expenditure. The analysis uses Convolutional Neural Networks to perform deep learning regression on each accelerometer configuration. The performance of this method is benchmarked against a set of conventional supervised machine learning and simplified regression models. Based on a leave-one-subject-out cross-validation, the results show that deep learning can achieve a comparable performance to the best conventional supervised learning, and significantly outperformed the simplified regression approaches.