Research Project
9994897
Exercise is a challenge for people with type 1 diabetes (T1D) because different activity types and intensities have different impacts on glucose concentrations. This challenge is amplified on competition days, as the added stress and high-intensity activity causes drastically different glucose responses than during exercise on non- competition days. This can occur due to a combination of increased hepatic glycogenolysis anaerobic exercise and the counterregulatory hormone response associated with the stress of competition. Thus, many people with T1D limit their physical activity due to fear of hypoglycemia during or after the activity and the majority of people with T1D do not engage in enough physical activity with less than 20% performing aerobic exercise more than twice per week. Regardless of the desire to exercise, people with T1D may be caught in scenarios where they are doing physical activity such running to catch a bus to commute to work may cause a spike in blood glucose concentration due to the anaerobic nature of the sprint and the added stress of arriving late if he or she were to miss the bus. One would not have had time to consider the appropriate insulin dose to prevent hyperglycemia and would benefit greatly from a system that would monitor these states and make corrections without requiring user input. Artificial pancreas (AP) systems will simplify the treatment of T1D and more advanced research is needed to increase their efficacy during and after exercise. The AP automates the computation of the proper insulin dose by using a continuous subcutaneous glucose sensor (CGM), an automatic control system, and an insulin pump that continuously infuses insulin into the subcutaneous tissue. The limitation of this configuration is its reliance on a single measurement, the glucose concentration, which varies in response to changes in many factors such as meals and exercise. The Cinar group at Illinois Institute of Technology pioneered the multivariable AP development. Signals streamed in real time from wearable devices complement glucose data to build accurate models to predict future variations in glucose levels. These models are updated recursively with each new set of measurements and used in hypoglycemia prediction and adaptive control. A new generation multivariable AP system that will incorporate the presence of competition stress with knowledge of physical activity type and intensity with well-developed machine learning algorithms for data interpretation in real-time, hypoglycemia early-warning systems, and automatic control will be more effective in improving glucose control in physically-active people with T1D. Additionally, this technology will be safer by dramatically reducing the number and duration of hypoglycemic and hyperglycemic events, as compared to the current guidelines for maintaining euglycemia in exercise. Such AP systems can only be developed by using a sophisticated multivariable approach that includes glucose concentrations and a number of other physiological variables that impact glucose homeostasis during physical activity and under the presence of competition stress.
