In this project, we will evaluate the efficacy of a novel approach to tailoring behavioral interventions for self-management of type 2 diabetes to individuals' behavioral and glycemic profiles discovered using computational learning and self-monitoring data. Growing evidence suggests significant differences in individuals' physiology and glycemic function, and their cultural, social, and economical circumstances that impact diabetes self-management. These discoveries highlight the need for personally tailoring both medical treatment and behavioral interventions. Yet tailored behavioral interventions proposed thus far typically focus on motivation for behavior change and individuals' psycho-social characteristics, rather than personalizing self-management strategies, such as changes in diet and physical activity. Moreover, tailoring typically relies on expert identification of tailoring variables and decision rules, and on standard surveys for assessment these variables. Data collected with self- monitoring can more accurately reflect an individual's behaviors and glycemic patterns, thus highlighting their ?behavioral phenotypes?, yet such data are rarely utilized in tailoring. The ongoing focus of this research is on developing informatics interventions for diabetes self- management, with a specific focus on personal discovery with self-monitoring data and on problem-solving for improving glycemic control. In the proposed research we will introduce GlucoType that relies on computational pattern analysis of data collected with self-monitoring technologies to identify behavioral patterns associated with poor glycemic control and formulate personalized behavioral goals for changing problematic behaviors. In our preliminary studies we have established that 1) computational phenotyping methods can accurately identify systematic associations between individuals' activities and changes in BG levels; 2) these patterns can be automatically translated into behavioral goals formulated in a natural language in a way consistent with goals formulated by diabetes experts, and 3) individuals with T2DM can understand and follow these behavioral goals and engage with GlucoType for personal self- management of diabetes. In the proposed research we will evaluate GlucoType's efficacy in a randomized controlled trial conducted with a practice-based research network (PBRN) of Federally Qualified Community Health Centers (FQHCs) in the metropolitan New York area.
The proposed informatics intervention will be designed to help individuals with diabetes improve their problem-solving abilities, develop better self-case behaviors and improve their health. As such, this project has a high potential public health impact.
