Understanding the Anatomy of Virtual Coaches - A Step Towards the Model-Driven Development of Digital Therapeutics

Virtual coaches (VCs) hold the promise to deliver personalized and data-driven care directly to patients. They are considered the vision of digital therapeutics by embodying the notion of an anytime-available pocket companion. While existing literature explores various aspects of VCs, comprehensive conceptual analysis is lacking to guide their model-driven development and integration. Model-driven approaches aim to streamline the technical process and adaptation to new disease scenarios, thereby accelerating the time to patient benefit and enhancing software quality. The present paper proposes foundational design principles of VCs and their detailed configuration possibilities, laying the groundwork for domain-specific modeling languages to be instantiated in low-/no-code platforms. Based on a qualitative content analysis of empirical work, the proposed design principles reflect the “voice of the developers” and can provide conceptual guidance for researchers and practitioners alike. Further, the analysis results reveal underexplored areas and opportunities for future research.