Incorporating an Exercise Detection, Grading, and Hormone Dosing Algorithm Into the Artificial Pancreas Using...

Area of Science:

• Biomedical Engineering
• Endocrinology
• Diabetes Technology

Background:

• Artificial endocrine pancreas (AP) systems aim to automate diabetes management.
• Exercise poses a significant challenge to glucose control in type 1 diabetes.
• Accurate exercise detection is crucial for effective AP system response.

Purpose of the Study:

• To develop and validate methods for automatic exercise detection for AP systems.
• To improve glucose level estimation during exercise within AP models.
• To create and evaluate an exercise-specific insulin dosing adjustment algorithm.

Main Methods:

• Utilized accelerometer and heart rate data from 13 type 1 diabetes subjects during an inpatient study.
• Employed a validated regression model for exercise event detection.
• Incorporated an exercise model into an in silico glucoregulatory model.
• Developed and tested an exercise dosing adjustment algorithm using in silico simulations.

Main Results:

• Achieved 97.2% sensitivity and 99.5% specificity in detecting exercise events.
• Demonstrated that including an exercise model improves glucose drop estimation during physical activity.
• Validated the performance of the exercise dosing adjustment algorithm in silico.

Conclusions:

• Automatic exercise detection using wearable sensors is feasible and accurate for AP systems.
• Enhanced glucoregulatory models with exercise components improve glucose prediction.
• An exercise dosing adjustment algorithm can aid in managing glucose levels during physical activity for type 1 diabetes.