Design and formative usability evaluation of a prototype integrated digital health platform for chronic care support

Chronic care management platforms often support only one or two core pillars of care, such as patient monitoring, patient activation, and data-driven care coordination. This fragmentation can limit the continuity of support and reduce the usefulness of shared patient data among patients, coaches, and clinicians.

This study describes the design, software architecture, and formative usability evaluation of the Digital Health Navigator (DiNa) platform, an integrated system designed to bring together the core pillars of chronic care management. DiNa combines wearable and self-reported monitoring, a large language model-powered embodied conversational agent, and a coordinator dashboard within a single modular ecosystem. DiNa was implemented as an Android application, a web-based coordinator dashboard, a microservices backend, and wearable integrations with a smartwatch and a continuous glucose monitor. The embodied conversational agent supports goal setting, logging, reminders, and knowledge-grounded education through coaching informed by motivational interviewing and cognitive behavioural therapy. We conducted moderated laboratory think-aloud usability sessions with 10 relatively young healthy volunteers familiar with digital health tracking. Participants explored the application and completed structured tasks, including setting goals, logging meals with an image, reviewing history, interacting with the coach, and viewing trends.

Outcomes included task performance, the System Usability Scale (SUS), a brief NASA Task Load Index (NASA-TLX), three acceptability items, and a reflexive thematic analysis of interviews. Participants completed the core tasks in a facilitated session, and a small number of critical breakdowns were observed and quantified. Mean SUS was 76.1 (SD 15.0), indicating generally good usability. Overall, the NASA-TLX workload averaged 26.7, with mental demand and effort higher than the other subscales and the remaining subscales low. Acceptability ratings indicated moderate early acceptability, with perceived usefulness rated more positively than near-term use or recommendation. Qualitative analysis identified five themes, including burdensome and sometimes unclear logging workflows, uneven goal setting, a friendly but occasionally generic coach, interface elements that reduced confidence, and a strong interest in integration with existing devices and applications.

These findings provide early evidence that an integrated platform combining continuous sensing, conversational coaching, and clinician-facing coordination is technically feasible and usable in a controlled setting with healthy volunteers, and they identify the workflow refinements needed before evaluation in clinically representative populations.