Health recommender systems for behavior change: exploring their potential for smoking cessation

Recommender systems are artificial intelligence algorithms that can identify the most relevant piece of content for their users. This thesis analyzed the state of the art of these health recommender systems for smoking cessation and identified that the previous implementations were rather descriptive, theoretical, did not explain how such systems were created, and did not use any Health promotion theoretical factors and behavior change theories. Therefore, this thesis designed a health recommender system to support smoking cessation grounded in behavioral science, which was tested in a 6-month trial. The system used a mobile app to deliver personalized motivational messages to the smokers, who could feedback the system to make it learn about their preferences and improve the message personalization over time. The trial compared this system with a simplified version for message smoking abstinence, appreciation, and engagement.