Designing and Testing HealthTracker for Activity Recognition and Energy Expenditure Estimation within the DAPHNE Platform

This paper describes the design and evaluation of a mobile software library, HealthTracker, which aims to produce activity and energy expenditure estimations in real-time from accelerometer and gyroscope data provided by wearable sensors. Using feature extraction together with a classifier trained using machine learning, the system will automatically and periodically send all the produced estimations to a cloud-based platform that will allow later evaluation by both the user and a physician or caretaker. The system is presented within the DAPHNE platform, an ICT ecosystem designed to provide a means for remote health and lifestyle monitoring and guidance between physicians and their patients.