Probabilistic Meal Detection and Estimation in Type 1 Diabetes With Extreme Shape Variability

Area of Science:

• Biomedical Engineering
• Computational Physiology
• Diabetes Technology

Background:

• Particle filters are conceptualized for meal detection and shape estimation in type 1 diabetes (T1D).
• This approach aims to improve meal announcements for T1D patients.
• Leverages continuous glucose monitoring (CGM) and insulin pump data, enhancing prior accelerometer-based methods.

Purpose of the Study:

• To develop and evaluate an algorithm for accurate meal detection and size estimation in T1D.
• To enhance glucose management through improved meal event recognition.
• To provide data for closed-loop insulin delivery systems.

Main Methods:

• Utilizes glucose appearance rate (RA) curves from triple-tracer studies.
• Normalizes curves using population meal size distributions (NHANES) and integrates insulin response data.
• Individualizes scaling based on insulin sensitivity and carbohydrate-to-insulin ratios.

Main Results:

• Preliminary findings indicate successful meal detection using the Tidepool dataset.
• The developed algorithm demonstrates effectiveness in identifying meal events.

Conclusions:

• The approach offers potential for better blood glucose management in T1D.
• Provides crucial information on glucose response and meal sizes for closed-loop control.
• Future work includes algorithm enhancement and integration of additional data types.