An Agent Program in an IoT System to Recommend Plans of Activities to Minimize Childhood Obesity

Abstract

Overweight and obesity in children is a recognized worldwide epidemic. They are associated with several current and future chronic diseases. OCARIoT is a joint EU-Brazil joint that aims to develop a sophisticated, noninvasive, unobtrusive, personalized IoT system to detect and normalize the behaviors that put a child at risk of developing obesity or eating disorders. In a recent written work, we proposed the design of an agent-based approach to recommend individual physical and food-related activities, based on data collected from wearable devices. In this paper, we present the design of an expanded approach that, in addition to recommendations for individual activities, should recommend activity plans, i.e., sequences of activities organized to minimize childhood obesity. The first results with the extended version were very promising. During the experiments, the selected individual activities and sequences of activities organized by the approach proved to be effective in conducting children, with different profiles and initial states, to the desired states of various attributes associated with childhood obesity.