Unhealthy behaviors, including physical inactivity, poor diet, tobacco use, and alcohol misuse, continue to be significant contributors to global mortality, highlighting the pressing need for effective, scalable interventions to promote sustained health behavior change. In recent years, mobile health (mHealth) applications have emerged as promising tools for promoting such change, offering cost-effective, accessible, and personalized solutions. However, despite their potential, many mHealth applications still adopt a "one-size-fits-all" approach, which fails to account for the considerable variability in user characteristics that influence engagement and behavioral outcomes.
This thesis addresses the challenge of personalization in mHealth by proposing a structured, empirically validated framework that links user profiles to mechanisms. The work is grounded in established behavior change theories and gamification principles. It identifies key dimensions relevant to personalization, including personality traits (e.g., Big Five), player types (e.g., Hexad, BrainHex), and demographic variables (e.g., age, gender). A multi-stage research process was conducted. An initial scoping review established a comprehensive preference matrix linking user types to mechanisms. This matrix was subsequently validated through an empirical study involving self-reported profiling and mechanism selection tasks.
In order to operationalize these findings, the thesis introduces two major contributions. First, it proposes a preference matrix that systematically maps user characteristics to mechanisms, providing practical guidance for developers and researchers aiming to design adaptive mHealth interventions. Secondly, it presents a formal ontology that encodes these relationships in a semantic framework, thereby enabling advanced querying, reuse, and future integration with broader behavior change ontologies.
The findings suggest a substantial degree of convergence between literature-derived predictions and empirical user preferences, thereby substantiating the external validity of the proposed framework. Furthermore, the work delves into the methodological and ethical considerations associated with user profiling, advocating for the utilization of validated instruments over opaque, automated inference systems. The limitations of the study are acknowledged, including the exclusion of some dimensions, such as message framing and social norms, which are proposed as avenues for future research.
In summary, this thesis presents a principled, extensible approach to user-centered personalization in mHealth for behavior change. The integration of theoretical models, empirical data, and semantic tools is a significant contribution to the development of more engaging, effective, and individualized mhealth interventions.
