Smart Selection of Useful Insights from Wearables

Abstract

The popularity of wearable-devices equipped with inertial measurement units (IMUs) and optical sensors has increased in recent years. These sensors provide valuable activity and heart-rate data that, when analysed across multiple users and over time, can offer profound insights into individual lifestyle habits. However, the high dimensionality of such data and user preference dynamics present significant challenges for mining useful insights. This paper proposes a novel approach that employs natural language processing to mine insights from wearable-data, utilising a neural network model that leverages end-to-end feedback from users. Results demonstrate that this approach effectively increased daily step counts among users, showcasing the potential of this method for optimising health and wellness outcomes.