Physical Activity Recommendation for Exergame Player Modeling using Machine Learning Approach

Abstract

Exergames are effective tools to motivate and promote daily physical activities. However, previous studies indicated that many people who start any type of exercise drop out of the program before establishing new habits. Research has shown that personalization is key to effective game-based interventions. Player modeling and recommender systems are used for personalizing contents and services in many applications. In exergames, we believe it is important to continuously recommend personalized and appropriate types of physical activity and contents in order to improve the effectiveness of the game. In this paper, we proposed and validated the design of a personalized physical activity recommender system for exergames based on a study of participant's preferred activities. The proposed approach resulted in more accurate recommendations when comparing to an existing model in predicting users' preference toward physical activity types.