Weight loss and low-carbohydrate intake scenarios in a predictive model reduced 1-year HbA1c by 0.2% compared to weight gain and high intake among youth with type 2 diabetes.
Observational (n=2,254)
Yes
Can a predictive model of HbA1c trajectories accurately generate patient-matched synthetic controls for youth-onset type 2 diabetes?
A nonlinear mixed effects model accurately predicts HbA1c progression in youth-onset type 2 diabetes, enabling the creation of synthetic controls for future trials and individualized management.
p-value: p=<0.01
Introduction and Objective: Youth-onset type 2 diabetes (T2D) is highly heterogeneous, complicating trial evaluation and requiring individualized management. We developed a predictive model of HbA1c trajectories to generate patient-matched synthetic controls (digital twins) using routine clinical features. Methods: We modeled 12,970 HbA1c measurements over a median 4.5 yr from 699 youth (median age 14 yr) in the TODAY study using a nonlinear mixed effects approach. The model was used to simulate HbA1c trajectories for 1,555 real-world youth from the SEARCH study and the UC Health Data Warehouse using patient-level baseline data under weight and dietary scenarios. Results: A progression model on HbA1c formation best described the data (individual-level fit R2=0.81). Elevated triglycerides, lower AST/ALT ratio, longer T2D duration, and Tanner stage 1-3 at baseline, plus higher BMI Z-score and carbohydrate (carb) intake during treatment, predicted rapid HbA1c progression (P0.01). Simulations showed distinct progression by baseline risk (median 1-yr HbA1c 5.4-7.5%), with weight loss and low-carb intake reducing 1-yr HbA1c by 0.2% vs. weight gain and high intake. Conclusion: Our HbA1c progression model in youth has the potential to create synthetic controls for trials and support risk-stratified strategies from diagnosis. It will be prospectively evaluated as a digital twin intervention in a clinical trial to project individual trajectories and guide behavioral changes. Disclosure E. Yang: None. S. Hwang: None. X. Luu: None. M.L. Tanenbaum: Speaker's Bureau; Current; Beta Bionics, Inc. Other - Honoraria; Ended; Sanofi. P. Phillips: None. R. Savic: None. S. Srinivasan: Research Support; Current; Abbott, Eli Lilly and Company. Funding American Diabetes Association (7-24-ICTST2DY-05)
Yang et al. (Fri,) conducted a observational in Youth-onset type 2 diabetes (n=2,254). Weight loss and low-carbohydrate intake vs. Weight gain and high carbohydrate intake was evaluated on HbA1c progression (p=<0.01). Weight loss and low-carbohydrate intake scenarios in a predictive model reduced 1-year HbA1c by 0.2% compared to weight gain and high intake among youth with type 2 diabetes.