Machine learning techniques in eating behavior e-coaching

摘要: The rise of internet and mobile technologies (such as smartphones) provide a harness data an opportunity to learn about peoples' states, behavior, context in regard several application areas such health. Eating behavior is area that can benefit from the development effective e-coaching applications which utilize psychological theories science techniques. In this paper, we propose framework how machine learning techniques effectively be used order fully exploit collected ("Think Slim") designed assess eating using experience sampling methods. overall goal analyze individual states person status (emotions, location, activity, etc.) their impact on unhealthy eating. Building different participants, classification algorithm (decision tree tailored longitudinal data) warn people prior possible event clustering (hierarchical agglomerative clustering) for profiling participants generalize new users application. Finally, offer feedback via adaptive messages (intervention) recommendations events presented. Results applying our methods reveal clustered six robust groups based there are specific rules discriminate conditions lead healthy versus Consequently, these utilized semi-tailored who, through method, assisted under more prone Effectiveness approach confirmed by observing decreasing trend rule activation towards end intervention period.