mHealth App recommendation based on the prediction of suitable behavior change techniques

In light of individuals' increasing concern regarding their physical health, mobile health applications (mHealth Apps) have gained popularity in recent years as important tools for addressing health problems. However, users find it challenging to choose appropriate mHealth Apps, as these Apps incorporate diverse behavior change techniques (BCTs), and their individual behavioral intervention effects on users vary. This study proposes a novel BCT-based mHealth App recommendation method to suggest suitable mHealth Apps to users. Specifically, we encode mHealth Apps to obtain information on the BCT adopted by the Apps. Based on the combination of BCTs in each mHealth App and its usage information, we construct a User-BCT matrix to represent users' preferences concerning BCTs. We also construct a user profile for each user, which considers their characteristics related to BCTs. Next, we build a prediction model that links each user's profile to BCTs, and use the AdaBoost algorithm to predict suitable BCTs for a target user. Finally, we recommend mHealth Apps with the highest BCT-matching levels to a target user. We also investigate the performance of the proposed method using a real dataset. The experimental results demonstrate the advantages of the proposed method.