The design and evaluation of a simulation-based behavior change intervention for individuals with type 2 diabetes

This dissertation describes a line of research that addresses translational research questions related to the use of computerized simulation to affect the knowledge, beliefs, motivation and self-management behaviors of individuals with chronic disease. The specific research projects focus on type 2 diabetes (T2DM) and physical activity as exemplars of a prevalent chronic disease and an underutilized selfmanagement behavior, respectively. We first describe a conceptual framework for the design of Consumer Health Informatics (CHI) applications. The design of an envisioned diabetes self-management application is described as an example of the application of design principles derived from this framework. Subsequent chapters describe tests of research questions related to this envisioned intervention. The second chapter describes the development and preliminary evaluation of the interface for the intervention described above. The estimation of simulated glucose curves for individuals with T2DM is described. Next, the formative evaluation of a paperbased prototype based on those curves and a novel method to measure individuals' outcome expectations are described. The third chapter describes a randomized experiment of a narrated simulation based on simulated glucose curves. This trial tested the question: can computerized simulations affect the beliefs and behaviors of individuals with T2DM? In this experiment participants' beliefs changed in accordance with the discrepancy between the presented evidence, and their prior beliefs, and in combination with the completion of a planning intervention, which resulted in significantly greater increase in physical activity The fourth chapter describes a test of the question: can predictive models of the acute physiologic effects of behavior be individualized? In this study we compared different predictive modeling techniques and found that a mixed effects modeling approach improves in accuracy as the individual contributes more data. This result is foundational to the development of the next generation of our simulation-based intervention, and has implications for CHI as a field; these are discussed. The dissertation concludes with a review of the strengths and limitations of the work described, a discussion of the implications of this work for consumer health informatics and a brief discussion of the next steps in this line of research.