Modeling Glucagon Action in Patients With Type 1 Diabetes

Area of Science:

• Biomedical Engineering
• Endocrinology
• Computational Biology

Background:

• Artificial pancreas (AP) technology aims to improve type 1 diabetes (T1D) management through automated glucose control.
• Current AP simulation models often lack comprehensive glucagon action submodels, hindering algorithm optimization.
• Developing accurate glucagon action models is crucial for advancing AP systems.

Purpose of the Study:

• To evaluate and compare nine distinct candidate models of glucagon action for use in T1D artificial pancreas simulations.
• To identify the most accurate mathematical representation of glucagon's effect on glucose production.

Main Methods:

• Utilized experimental data from eight T1D subjects receiving subcutaneous glucagon boluses, measuring plasma insulin, glucagon, and endogenous glucose production.
• Employed a Bayesian approach for parameter estimation of nine candidate glucagon action models.
• Assessed model performance using the deviance information criterion (DIC) for model comparison.

Main Results:

• Identified a best-fit model incorporating insulin-dependent suppression of glucagon action.
• The optimal model also accounted for the effects of both glucagon concentration and its rate of change on glucose production.
• This model demonstrated superior accuracy in simulating glucagon's physiological impact compared to other candidates.

Conclusions:

• The developed glucagon action model provides a more accurate representation for AP simulations.
• Accurate glucagon modeling is essential for optimizing AP dosing algorithms and improving T1D treatment.
• This research contributes to the advancement of closed-loop artificial pancreas systems.