A Data Science and Machine Learning Approach to Measure and Monitor Physical Activity in Children

Abstract

Physical Activity is a fundamental component for the maintenance of a healthy lifestyle. Recommendations for physical activity levels are issued by most governments as part of public health measures. Therefore, it is vital for regulatory purposes, that there are reliable measurements of physical activity. However, the techniques and protocols used in existing physical activity research, including laboratory-based measurement, have received increasingly critical scrutiny in recent times. Consequently, physical activity researchers have begun to explore the use of wearable sensing technology to capture large amounts of data and the use of machine learning techniques, specifically artificial neural networks, to produce classifications for specific physical activity events. This paper explores this idea further and presents a supervised machine learning approach that utilises data obtained from accelerometer sensors worn by children in free-living environments. The paper posits a rigorous data science approach that presents a set of activities and features suitable for measuring physical activity in children in free-living environments. A Multilayer Perceptron neural network is used to classify physical activities by activity type, using ecologically valid data from body worn accelerometer sensors. A rigorous reproducible data science methodology is presented for subsequent use in physical activity research. Our results show that it was possible to obtain an overall accuracy of 92% using the initial data set, and 99.8% using interpolated cases.