Advanced Insulin Bolus Advisor Based on Run-To-Run Control and Case-Based Reasoning

Area of Science:

• Biomedical Engineering
• Computational Biology
• Endocrinology

Background:

• Diabetes management requires precise insulin dosing, often challenging with standard bolus calculators.
• Multiple daily injections and insulin pump therapy necessitate adaptable tools for optimizing glycemic control.
• Continuous glucose monitoring (CGM) data offers potential for more personalized insulin therapy adjustments.

Purpose of the Study:

• To develop and validate an advanced insulin bolus advisor for diabetes patients.
• To enhance the adaptability and flexibility of standard bolus calculators using novel algorithms.
• To improve glycemic outcomes, including reducing mean blood glucose and eliminating hypoglycemia.

Main Methods:

• A novel algorithm combining retrospective run-to-run (R2R) optimization with case-based reasoning (CBR) was developed.
• The system was integrated into a smartphone application for user accessibility.
• In-silico studies were conducted using the FDA-accepted UVa-Padova type 1 diabetes simulator, with updated scenarios for real-world variability.

Main Results:

• The CBR(R2R) algorithm significantly reduced mean blood glucose levels in simulated adult and adolescent populations.
• The system demonstrated a notable increase in time spent within the euglycemic range.
• Hypoglycemia was completely eliminated in simulations using the CBR(R2R) algorithm, outperforming the standalone R2R approach.

Conclusions:

• The developed CBR(R2R) algorithm represents a significant advancement in insulin bolus advisory systems for diabetes.
• Integrating case-based reasoning with run-to-run optimization enhances therapeutic outcomes and patient safety.
• The smartphone-based system shows promise for improving diabetes management in real-world settings.