The FiberMore generative AI app significantly increased dietary fiber intake by 8.2 g compared to control (P<0.001) in adults with Type 2 diabetes.
RCT (n=70)
Single-blind
randomized
Yes
Does a generative AI-enabled mobile app improve dietary fiber intake and HbA1c levels in adults with Type 2 diabetes?
A generative AI-powered mobile app significantly increased dietary fiber intake in adults with Type 2 diabetes, though it did not lead to a statistically significant reduction in HbA1c compared to control at 12 weeks.
Mean Difference: 8.2
Absolute Event Rate: 7.3% vs -0.9%
p-value: p=<0.001
Introduction and Objective: A high intake of dietary fiber has been shown to improve glycemic management in individuals living with Type 2 diabetes (T2D). We developed FiberMore, a generative AI-based mobile app grounded in the Theory of Planned Behavior, to address inadequate fiber intake in individuals with T2D, and assessed its efficacy in increasing fiber intake and reducing HbA1c levels. Methods: This is a single-blinded randomized controlled trial targeting 70 adults with T2D from two hospitals in Tokyo with a 12-week intervention and 12-week observation. The intervention group used FiberMore, which employs a fine-tuned GPT-4o model to analyze meal nutrients and provides personalized, real-time feedback on fiber. An AI chatbot embedded in FiberMore delivers personalized goal setting, barrier/coping-strategy setting, emotional attitude assessment, and education. The control group received sham salt-reduction education and recorded daily salt-reduction efforts in a diary. Results: The intervention group showed a greater fiber increase from run-in to Week 12 than the control group (+7.3 g vs -0.9 g), with a significant between-group difference of +8.2 g (P 0.001) representing a 64% increase. HbA1c decreased by 0.35% in the intervention group and by 0.14% in the control group, resulting in a not-significant between-group difference of 0.21% after baseline adjustment (P = 0.170, 95% CI −0.52 to −0.092), which was consistent with prediction from a published fiber dose-response relationship (estimated −0.27%, 95% CI −0.55 to 0.00) for a 64% increase in fiber. Higher engagement with AI-delivered behavior change strategies, including goal setting, barrier/coping-strategy setting, and emotional attitude assessment, were significantly associated with greater dietary fiber intake (P 0.05). Conclusion: A generative AI-powered, app-based intervention led to a significant increase in dietary fiber intake and a corresponding reduction in HbA1c. Disclosure W. Sze: None. D. Lane: Stock/Shareholder; Current; WaShiLa Health. R. Nakada: None. K. Waki: None.
Sze et al. (Mon,) conducted a rct in Type 2 diabetes (n=70). FiberMore (generative AI-based mobile app) vs. Sham salt-reduction education and daily diary was evaluated on Change in dietary fiber intake from run-in to Week 12 (MD 8.2, p=<0.001). The FiberMore generative AI app significantly increased dietary fiber intake by 8.2 g compared to control (P<0.001) in adults with Type 2 diabetes.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: