Toward Enhancing Diabetes Self-Management with Personalization Through Human Digital Twins for Behavior Change

Patients with type 2 diabetes need to adhere to a healthy lifestyle to prevent suffering from short-term and long-term escalations. A key aspect of treatment is to supplement educational content with techniques for behavior change. Unfortunately, introducing change and maintaining it is notoriously hard. We explore a novel approach to personalization in the context of Human Digital Twins (HDTs), that facilitates a bidirectional data flow between patients and their digital counterparts. While current mHealth tools show potential, there is room for improvement in incorporating the concept of such twins and considering the complexity of behavior change. We developed a novel architecture that enables mHealth tools, through the lens of the Transtheoretical Model of behavior change, to focus on multiple stages of change and personalize functionalities based on an HDT model that is trained more holistically. Our modular HDT is made of three subcomponents: a health literacy model, a daily activities model, and a model of the user’s player type. Our primary contribution is the demonstration of its relevance and feasibility toward the application of personalization for behavior change in patients with diabetes.