Intelligent Decision Support in Personal Health: Personalized Health Coaching in Type 2 Diabetes

Chronic diseases, such as type 2 diabetes (T2D), hypertension, and obesity place an ever-increasing burden on individuals and society at large. Increased availability of self-monitoring technologies in health enabled a new class of personal informatics technologies that incorporate personal data to facilitate self-management through reflection and increased self-knowledge. However, these technologies often place burden of data analysis on individuals themselves, which may present barriers to low resource, low literacy populations. Furthermore, increased knowledge of patterns in the past may not be enough to suggest future action. New advances in machine learning and computational modeling can alleviate cognitive burdens of data analysis and identify meaningful trends and patterns in self-monitoring data. Furthermore, these techniques can be used to suggest future actions with potential health benefits. However, translating computational inferences into useful decision support in non-trivial. In this chapter, we describe our experience designing an intelligent interactive system for supporting self-management in type 2 diabetes. This system relies on machine learning to identify appropriate self-management goals using self-monitoring data and provide in-the-moment decision support using an interactive mHealth solution. We use our experiences to draw implications for future intelligent systems for personalized decision support in personal health.