Motivating People to Move More with Personalized Activity and Tip Recommendations: a Randomized Controlled Trial

Abstract

As current mobile health interventions often suffer from user dropout and a lack of motivation, this study investigates motivation of insufficiently active people when receiving personalized recommendations to move more. We developed a content-based recommender system, based on our two datasets for physical activities and tips to break sedentary behavior with descriptive attributes, implemented in an Android app. A longitudinal user study was conducted with 25 participants over eight weeks, following a randomized controlled trial study design in which the experimental group receives personalized recommendations based on the user's mood, preference history, and estimated current situation (e.g., work or free time), while the control group receives non-personalized, random recommendations. Our results show that rating feedback was on average .596 out of 5 stars higher, and motivation to move on average .929 out of 4 higher in the experimental group. Additionally, 53 participants initially started the study and there was lower dropout after eight weeks in the experimental group. As our results confirm that the experimental group had more motivation to move, we propose integrating our personalization algorithm in future mobile health interventions as a solution to the challenge of keeping users engaged and motivated.