Hypoglycemia risk with physical activity in type 1 diabetes: a data-driven approach

Area of Science:

• Endocrinology
• Metabolic Diseases
• Exercise Physiology

Background:

• Physical activity (PA) offers significant health advantages for individuals with type 1 diabetes (T1D).
• Exercise-induced hypoglycemia is a major concern that can discourage regular PA participation among those with T1D.
• Understanding PA's impact on hypoglycemia is crucial for promoting active lifestyles in the T1D population.

Purpose of the Study:

• To investigate the association between common physical activities (walking, running, cycling) and the risk of hypoglycemia in individuals with T1D.
• To identify key predictors of exercise-induced hypoglycemia in T1D patients.
• To provide data-driven insights for managing hypoglycemia risk during PA.

Main Methods:

• Utilized real-world data from the Tidepool Big Data Donation Project, including PA details, continuous glucose monitoring (CGM) data, and insulin doses from 50 T1D participants.
• Developed linear regression models for walking, running, and cycling to predict hypoglycemia risk (average glucose deviation from 70 mg/dL) in the 2 hours post-exercise.
• Included predictors such as PA duration and intensity, pre-exercise glucose levels (adjusted AUC and start value), and recent bolus insulin doses.

Main Results:

• The glucose value at the start of exercise and the pre-exercise glucose adjusted AUC were highly significant predictors of hypoglycemia risk across all three activities (p < 0.001).
• PA duration, intensity, and 2-hour pre-exercise bolus insulin showed only weak associations with hypoglycemia risk for walking, running, and cycling.
• The models effectively predicted hypoglycemia risk based on initial glucose status and trends.

Conclusions:

• Initial glucose levels and recent glucose trends are the most critical factors influencing hypoglycemia risk during PA in individuals with T1D.
• While PA characteristics and insulin doses have some influence, they are less predictive than baseline glycemic status.
• These findings support a data-driven approach for T1D individuals to prepare for PA and mitigate hypoglycemia risks.