Multivariable Artificial Pancreas: Detecting and Mitigating Unannounced Physical Activity and Acute Psychological Stress

The development of an artificial pancreas (AP) has gained considerable progress over the last decade. The goal of AP systems is to develop a device that will provide insulin on demand in response to blood glucose levels in the same manner as the own pancreas. The first generation of AP systems, the hybrid closed-loop AP, collects data from continuous glucose monitoring (CGM) devices and relies on manual user inputs for mitigating the effects of meals and physical activities. The objective of this proposal is to demonstrate a viable, functionally integrated multivariable artificial pancreas (mvAP) that will address meal, physical activity (PA) and acute psychological stress (APS) challenges without any manual inputs to better regulate glucose levels of people with diabetes. Acute psychological stress and many other forms of PA besides planned exercise can affect blood glucose levels and cause challenges to maintaining euglycemia for people with type 1 diabetes mellitus (T1DM). Various PA and APS affect the metabolism and sensitivity to insulin in different ways. Hence, their types, intensities and durations, and their individual and concurrent presence must be detected in order to determine the optimal insulin administration. The mvAP approach provides a well-integrated and user-friendly technology with minimal burden on the user and mitigates the effects of unexpected PA and APS inducements. The objective of this proposal is to demonstrate a viable, functionally integrated mvAP that will address meal, PA and APS challenges without any manual inputs to better regulate glucose levels of people with T1DM. Our hypothesis is that a new generation multivariable AP that incorporates real-time detection and determination of the characteristics of PA and APS and mitigation of their effects by automatic control will be more effective in improving glucose control in people with T1DM. Additionally, this technology will be safer compared to APs based exclusively on CGM data, by reducing the number and duration of hypoglycemic and hyperglycemic events. Such mvAP systems can only be developed by using a sophisticated multivariable approach that uses real-time information from CGMs and physiological variables obtained from wearable devices worn in free-living daily life. It is proposed to build on this framework a fully automated mvAP technology that will mitigate meal, PA and APS challenges, including unexpected and unplanned events. The specific aims of the proposed research are: Aim 1. To develop the mvAP algorithms and modules that identify various types of physical activities, acute psychological stress episodes, their concurrent presence and their characteristics in real time. Aim 2a: To conduct open-loop studies in clinic and in free living to expand the types and intensities of PA and APS inducements, to enrich our database with data collected during activities of daily living (alone or coupled with APS), including unplanned spontaneous PA and APS events. Aim 2b:To extend our multivariable glucose-insulin-physiological variables simulator (mGIPsim) for in silico studies of the mvAP for simulating APS inducements and their effects on glucose levels and physiological variable outputs of the simulator. Aim 3: To conduct clinical experiments with the second generation mvAP in clinical settings and in free living to assess the performance of our fully-automated mvAP in closed-loop operation. At this time, IRB approval for Aim 1, 2a and 2b has been obtained. Aim 3 will be conducted at a later date. The clinical studies will take place at the University of Illinois at Chicago (UIC)-College of Nursing and at the subjects' home. The protocol will include 1 screening visit and 5 sessions at the laboratory. Subjects will wear a CGM and both Empatica and Actigraph wristbands throughout the study. The primary activities at each meeting will include: (1) screening; (2)estimation of maximal exercise capacity from sub-maximal exercise tests (bicycle and treadmill); (3) estimation of maximal strength from sub-maximal strength tests; (4) sub-maximal bouts of aerobic and resistance exercise; (5) use of Socially Evaluated Cold Pressor Test; (6) driving simulation; (7) resistance exercise, and activities of daily living with and without stress (e.g., mental calculations, video games); (8) use of GoPro/Dash Cam during driving (if appropriate) and questionnaires.These activities will be included in visit 3, 4 and 5 as appropriate. In addition, subjects will perform activities at home including: housekeeping chores, stationary bike (if available); treadmill (if available); walking; and light weights (if available) and a sleep monitor (Z-Machine Insight+). Periodically, the research assistant will call the subject during these times and ask them to perform stress-inducing activities while performing the activity. The stress inducing activities will include mental challenges such as a mathematical computation while performing the activity. The subjects will be called at home 3-5 times during the study. The fully automated algorithm will be tested in a home setting. Once the methodology is refined, Institutional Review Board (IRB) approval for the Aim 3 will be obtained.