Designing a Carbohydrate Counting App for Young Adults With Type 1 Diabetes: Usability Testing Interview Study

Carbohydrate counting (CC) assists people with type 1 diabetes (T1D) adjust mealtime insulin doses; however, it is often burdensome. Mobile apps can simplify this process by automating carbohydrate estimation and insulin calculations, yet no comprehensive solution currently combines photo-based carbohydrate recognition with an integrated bolus calculator.

This study aimed to identify user-informed design principles from usability testing interviews to optimize a novel app supporting young adults with T1D in CC and insulin dosing.

We conducted 4 iterative rounds of usability testing interviews, each with 3 to 5 participants, using a think-aloud protocol to evaluate how easily and effectively users interacted with the app and to identify areas for improvement. Interviews were analyzed qualitatively to derive main design principles, and findings from each round informed the refinement of the app prior to subsequent testing.

A total of 18 participants completed the usability testing (median age of 23, IQR 19-24 y and diabetes duration of 9, IQR 6-12 y; n=12, 66.7% young women). Thematic analysis highlighted that a person-centered design that prioritizes the lived experiences of youth with T1D was essential to position the app as a self-management support system, beyond a clinical tool. Personalization was central, including customizable treatment profiles, tailored dashboard metrics, esthetic preferences, and artificial intelligence-driven recommendations based on personal trends. Early usability barriers revealed the need for intuitive navigation, streamlined multistep processes, and clear guidance for data entry and interpretation. Participants valued culturally inclusive content and familiar terminology to enhance accessibility and engagement. Users perceived strong potential for the app to centralize diabetes management tasks, integrate contextual factors (eg, exercise, diet, and timing of insulin) with glucose data, generate sharable reports to facilitate patient-practitioner communication, and strengthen self-efficacy through personalized trend analysis. Concerns about over-reliance on automation underscored the necessity of transparent data verification and user override options to maintain trust in insulin dosing decisions.

Iterative usability testing highlighted the importance of balancing automation with user control, personalization, and contextual understanding of personal trends, as key design principles to enhance engagement and the apps' relevance as a self-management tool. Incorporating these features into a CC and insulin-dosing app could improve self-efficacy in youth living with T1D.