Multiple-Input Subject-Specific Modeling of Plasma Glucose Concentration for Feedforward Control.

Area of Science:

• Biomedical Engineering
• Control Systems Engineering
• Diabetes Technology

Background:

• Accurate blood glucose concentration (BGC) modeling is crucial for managing insulin-dependent diabetes.
• Type 1 diabetics (T1Ds) require robust control systems to manage BGC fluctuations from food, activity, and stress.
• Existing models may struggle with large datasets and real-world variability.

Purpose of the Study:

• To present a novel, subject-specific, multiple-input modeling method for BGC.
• To develop a stable model with strong causation attributes that prevents overfitting.
• To enable effective long-term feedforward control (FFC) for artificial pancreas systems.

Main Methods:

• Utilized a Wiener block-oriented methodology for BGC modeling.
• Focused on a free-living, outpatient, multiple-input approach.
• Emphasized stability and resistance to overfitting.

Main Results:

• The proposed method demonstrates strong causation attributes for BGC modeling.
• The model is stable and guards against overfitting, suitable for large input sets.
• The approach provides an effective modeling strategy for feedforward control.

Conclusions:

• The developed Wiener block-oriented modeling method offers a significant advancement for BGC control.
• This approach can enhance the performance of artificial pancreas systems for T1Ds.
• The modeling technique is well-suited for real-world, dynamic conditions and long-term FFC.