Introduction and Objective: AI-driven glucose predictions help CGM users manage glucose levels more proactively. The Accu-Chek® SmartGuide Predict app notifies of low glucose levels within 30 minutes (LGP), nighttime hypoglycemia risk (NLP), and provides a continuous 2-hour glucose forecast (GP). This study evaluates the potential glycemic benefits of the app via in-silico analysis.Methods: Using UVa’s digital twin methodology, 2458 days from 20 individuals in the REPLACE-BG dataset were simulated. Interventions: 15g fast-acting carbs after LGP notifications, 30g slow-absorbing carbs for high NLP risk at 22:30, and advancing/adding correction boli if GP forecast >220 mg/dL. Baseline intervention used a GP-like hyperglycemia mitigation strategy, substituting predictions with current CGM data. All hypoglycemia interventions are aligned with ADA guidelines. Nighttime meals/boli were omitted to better assess NLP.Results: LGP reduced %time <70mg/dL (4.2±1.5 vs 0.46±0.28 p<1e-4). GP reduced %time >180mg/dL (25.3±10.2 vs 22.0±8.6 p<1e-4). NLP decreased overnight %time <70mg/dL (5.6±2.2 vs 3.8±1.2 p<1e-4). Combining all features significantly reduced time outside range.Conclusion: The Accu-Chek® SmartGuide Predict app has the potential to enhance glycemic control in T1D. Limitations include the in-silico nature of the study and assumptions made in the user behaviour models.
P. Herrero: Employee; Roche Diabetes Care. M. Andorrà: Employee; Roche Diagnostics. Stock/Shareholder; Roche Diagnostics. M.D. Breton: Speaker’s Bureau; Sinocare Inc, Tandem Diabetes Care, Inc. Consultant; Roche Diabetes Care, Boydsense. Y. Klopfenstein: Consultant; Roche Diagnostics. T. Glatzer: Employee; Roche Diabetes Care. Stock/Shareholder; Roche Diabetes Care, Abbott, Dexcom, Inc.
The research has been funded by Roche Diabetes Care GmbH
Source link
Leave a Reply