JMIR Publications

ABSTRACT

Background:

Diabetes mellitus type 2 (T2D) requires continuous self-management beyond clinical consultations, particularly in lifestyle-related domains such as nutrition and physical activity. However, maintaining these lifestyle behaviors in everyday life remains challenging for most patients.

Objective:

This study aimed to evaluate DALIA, a conversational agent-based lifestyle monitoring for capturing lifestyle-related data from patients with T2D and supporting brief daily reflection. Specifically, we explored patients’ everyday experiences of T2D self-management (RQ1), assessed DALIA’s feasibility, usability, and acceptability for multidomain lifestyle data capture (RQ2), and examined its perceived clinical relevance to routine T2D care (RQ3).

Methods:

We conducted a five-day mixed methods pilot feasibility study with 4 patients with T2D and 6 healthcare professionals (HCPs): 2 general practitioners, 2 endocrinologists, and 2 certified diabetes educators. DALIA combined structured chats, questionnaires, daily goal reflections, step counts, and meal-image analysis. Baseline patient interviews explored self-management experiences, while DALIA logs and follow-up interviews assessed data capture, feasibility, usability, and acceptability. HCP interviews examined potential clinical relevance.

Results:

T2D Participants had largely integrated medical self-management into their daily routines, whereas lifestyle behaviors, particularly nutrition and physical activity, remained harder to manage consistently. During the five-day intervention, DALIA captured multidomain lifestyle data, including daily check-ins, physical activity, nutrition, stress, wellbeing, sleep, medication routines, and diabetes-related check-up information. Completion of the five lifestyle and wellbeing questionnaires was 100% (20 of 20), daily check-in completion was 95% (19 of 20), step count data availability was 70.8% (17 of 24), and meal-image data capture was 75% (9 of 12). Participants perceived DALIA as easy to use and understand, with a median System Usability Scale score of 80.0 (range 67.5-97.5). Daily prompts, the one-topic-per-day structure, and goal setting supported reflection on everyday behaviors. Limitations concerned notification timing, sensor reliability, and meal-image capture. HCPs considered the collected lifestyle information potentially relevant to clinical care for making everyday self-management more visible, identifying recurring behavioral patterns and exceptions, and supporting more focused consultations. Although general practitioners, endocrinologists, and certified diabetes educators differed somewhat in emphasis, all three groups saw potential value in DALIA as a complementary tool for T2D care.

Conclusions:

The findings support DALIA’s feasibility for short-term conversational lifestyle monitoring in patients with T2D and the potential usefulness of the resulting data for clinical discussions. Clinical use will require concise summaries, workflow integration, and technical robustness. Larger and longer-term studies should examine how DALIA-generated data can support patient-provider communication and sustained self-management.