ReplayBG

ReplayBG: Personalized Glucose–Insulin Dynamics Simulation Platform

ReplayBG constructs individualized glucose–insulin dynamic models from insulin dosages, carbohydrate intake, and continuous glucose monitoring (CGM) data, and performs retrospective in silico simulations of alternative therapeutic scenarios on the same historical dataset.

Key Features:

• Personalized Modeling: Estimates subject-specific glucose–insulin dynamics using real-world insulin, carbohydrate, and CGM data.
• In Silico Simulation: Simulates glucose concentration responses under modified insulin and carbohydrate treatment regimens.
• Model Validation: Validated using 100 virtual subjects from the UVa/Padova T1D Simulator (T1DS) across five insulin and carbohydrate modification scenarios, demonstrating high agreement with T1DS simulations.
• Real-World Case Evaluation: Applied to two clinical case studies to assess simulation performance in practical settings.

Scientific Applications:

• Therapy Evaluation in Type 1 Diabetes (T1D): Assesses potential effects of alternative insulin and carbohydrate treatment strategies on glucose dynamics prior to clinical trials.

Methodology:

ReplayBG operates in two stages: (1) model identification, which infers individualized glucose–insulin dynamics from insulin administration, carbohydrate intake, and CGM measurements; and (2) scenario simulation, which applies the identified model to retrospectively evaluate alternative insulin and carbohydrate regimens under controlled in silico conditions.