Alerting Type 2 Diabetics of Nocturnal Hypoglycemia Risk - Integrating user-centered design in a self-management app to present AI-based health predictions

Ottosen, Anders Solsbæk ; Andersen, Mads Sigh Lund ; Gustenhoff, Anne Schack

4. Term

Digitalisation and Application Development, MSc

2022

2022-05-31

29

Mennesker med diabetes, der bruger insulin, kan opleve hypoglykæmi—farligt lavt blodsukker. Det er særligt udfordrende om natten, hvor symptomer kan være svære at opdage. Computerbaserede forudsigelser kan advare på forhånd. I dette todelte studie brugte vi en brugercentreret designtilgang til at undersøge, hvordan en personlig forudsigelse af natlig hypoglykæmirisiko bedst kan formidles i en mobil selvhåndteringsapp (SMA) for personer med type 2-diabetes. Først gennemførte vi semistrukturerede interviews med potentielle brugere (N=10) og én endokrinolog for at forstå konteksten, afdække brugerbehov og finde foretrukne måder at vise risiko på. Derefter byggede vi, på baggrund af indsigterne, en prototype af en mobilapp og evaluerede den i tænke-højt-sessioner med kommende brugere (N=5), hvor deltagerne brugte appen og satte ord på deres tanker. Det hjalp os med at vurdere, om design og funktioner matchede behovene. På tværs af begge trin foretrak deltagerne tydelige farvesignaler og enkle visualiseringer suppleret med kort tekst eller tal. De efterspurgte også fleksibilitet: forskellige brugere ønsker forskellige detaljeringsgrader, og brugeren bør kunne tilpasse opsætningen af forudsigelsen. Vi identificerede derudover barrierer og drivkræfter for at tage mobile diabetes-SMA'er i brug og understreger behovet for samarbejde mellem brugere, sundhedsprofessionelle og forskere. På den baggrund giver vi anbefalinger til fremtidigt arbejde.

People with diabetes who use insulin can experience hypoglycemia—dangerously low blood glucose. This is especially challenging at night, when symptoms are harder to notice. Computer-based prediction methods can warn people ahead of time. In this two-part study, we used a user-centered design approach to learn how to communicate a personalized prediction of nighttime hypoglycemia risk in a mobile self-management app (SMA) for people with type 2 diabetes. First, we conducted semi-structured interviews with potential users (N=10) and one endocrinologist to understand the care context, identify user needs, and explore preferred ways to display risk. Then, based on these insights, we built a prototype mobile app and evaluated it in think-aloud sessions with prospective users (N=5), where participants used the app while verbalizing their thoughts. This helped us assess whether the design and functions matched user needs. Across both steps, participants favored clear color cues and simple visuals supported by short text or numbers. They also asked for flexibility: different users want different levels of detail, and they should be able to personalize how the prediction is configured. We further identified barriers and enablers for adopting mobile diabetes SMAs and emphasize the need for collaboration between users, health professionals, and researchers. We conclude with recommendations to guide future work.

[This summary has been rewritten with the help of AI based on the project's original abstract]

self-management app ; mHealth ; user-centered design ; AI prediction ; diabetes ; hypoglycemia

Download
View record in AAU Student Projects

Ottosen, Anders Solsbæk:

• Street level bureaucrats views on potentials and challenges with marginal jobs (2018)