Simulation-driven optimization of insulin therapy profiles in a commercial hybrid closed-loop system

Area of Science:

• Biomedical Engineering
• Endocrinology
• Computational Biology

Background:

• Automated insulin delivery (AID) systems have revolutionized type 1 diabetes (T1D) management, offering significant glucose control.
• Metabolic variability remains a challenge for hybrid closed-loop (HCL) systems, necessitating personalized insulin therapy profiles.
• The Control-IQ AID algorithm's performance is contingent upon optimized, customizable insulin profiles.

Purpose of the Study:

• To introduce and evaluate an Identification-Replay-Optimization (IRO) approach for gradually and safely optimizing insulin therapy profiles.
• To enhance the performance of the Control-IQ AID algorithm in managing T1D.
• To address metabolic variability challenges in HCL solutions.

Main Methods:

• Utilized the UVA/Padova T1D simulation platform with an adult cohort across diverse glycemic scenarios.
• Developed a personalized model of insulin-glucose dynamics for each subject based on daily records.
• Integrated identified models into a bi-weekly optimization procedure to adjust basal and bolus profiles, minimizing hypo- and hyperglycemia risks.
• Tested the strategy under various metabolic and behavioral variability scenarios.

Main Results:

• The IRO approach demonstrated improved glucose control over time in simulations.
• The strategy effectively mitigated risks associated with both hypoglycemia and hyperglycemia.
• Recommended insulin profile adjustments decreased with each cycle, indicating convergence under stable simulation conditions.

Conclusions:

• Simulation-driven tools that accurately replicate real-world data and predict changes can accelerate insulin therapy optimization.
• The IRO approach facilitates timely adjustments to insulin therapy in response to metabolic changes.
• Optimized insulin therapy through simulation leads to improved glucose control in T1D management.