2784-LB: Real-World Weight Loss Outcomes among Adults with Type 2 Diabetes Using an AI-Enabled Weight Management Mobile App

Introduction and Objective: Mobile app-based weight management programmes using artificial intelligence (AI) coaching are increasingly used to support lifestyle change in people with type 2 diabetes (T2D), but real-world outcomes remain limited. The aim of this study was to evaluate weight loss among adults with T2D using an AI-enabled weight management app, and to examine whether AI coaching chat engagement was associated with weight loss at 12 and 26 weeks. Methods: This retrospective observational study included adults with T2D enrolled in a digital weight management app. Weight loss (%) was assessed at 12 and 26 weeks. AI coach engagement was defined as mean days/week of chat use and categorised as 0, 1-3, or 4-7 days/week. The multivariable linear regression model was adjusted for age, sex, baseline BMI, and other in-app activity (days/week). Results: At 12 weeks (n=444), users were 73.2% female, mean age 54.4 (SD 14.6) years, BMI 36.2 (SD 6.9) kg/m2. Mean weight loss was 3.63 (SD 4.16) %. At 26 weeks (n=263), mean weight loss was 4.54 (SD 6.20) %. In unadjusted analyses, users engaging with AI chat 4-7 days/week achieved greater mean weight loss than those with lower engagement at both 12 weeks (4.69% vs 3.05% and 2.23%) and 26 weeks (6.70% vs 3.99% and 3.87%) (pairwise Mann-Whitney U with Holm-Bonferroni correction; adjusted p < 0.05). In the regression model adjusted for other weight loss predictors, AI chat engagement was no longer associated with weight loss. Conclusion: Adults with T2D achieved meaningful weight loss at 12 and 26 weeks using an AI-enabled weight management mobile app. Higher AI chat engagement was associated with greater weight loss in unadjusted analyses but not after adjustment for overall programme activity and other factors, suggesting AI chat reflects broader user engagement. Disclosure R. Huntriss: None. D. Nikogosov: None. R. Salimgaraev: None.