Insulin and carbohydrates are inputs for a mathematical model that need to predict glucose levels. Once a math model can predict glucose, then inputs can be varied to look at alternative realities (eg, impact of a smaller bolus). #ATTD2025

Professor Marc Breton presents on uses of simulations and digital twins for validating predictive AI systems to work with CGM data in #T1D. In silico subjects don't replicate the variable behaviour seen in vivo subjects. Work has focused on trying to replicate this variability using digital twins.