Page 309 - Glucose Monitoring Devices

316 CHAPTER 15 Automated closed-loop insulin delivery

constrained through predictions of the PIC obtained over the prediction horizon.
The ability of the adaptive MPC is demonstrated using the multivariable Glucose,
Insulin, and Physiological variable simulator (mGIPsim), which is based on a
modified Hovorka’s glucose-insulin dynamic model that takes into account the
effects of various physical activities. In addition to the CGM values, mGIPsim gen-
erates physiological variable signals reported by noninvasive wearable devices.
These physiological variables are used to evaluate the ability of the mAP system
and compare its performance to that of the conventional single input AP (sAP)
system. Aerobic medium-intensity exercises with treadmill and bicycle are used
for testing the mAP system. Twenty virtual subjects are simulated for 3 days
with varying times and quantities of meals consumed on each day and different
types and times of physical activities (Tables 15.1 and 15.2). The meal and phys-
ical activity information are not entered manually to the AP, as the AP controller is
designed to regulate the BGC in the presence of significant disturbances such as
unannounced meals and exercises. The metabolic equivalent of task (MET) values
computed by the simulator are used as physiological signals in the recursive system
identification technique. To show the efficacy of using physiological signals in the
AP system (the mAP case), the sAP case is also considered where no information
of physiological signals (MET values) are used in the AP system. The results based
on these two different cases (the mAP and sAP) are compared.
The evaluations of the closed-loop results based on the mAP and sAP are
presented in Figs. 15.4e15.7, and the quantitative metrics for comparing the sAP

Table 15.1 Meal scenario for 3-day closed-loop experiment using
mGIPsimdthe integrated multivariable metabolic and physiologic simulator.
First day Second day Third day
Meal Time Amount (g) Time Amount (g) Time Amount (g)
Breakfast 7:00 70 8:00 50 7:30 60
Lunch 12:00 60 12:30 80 13:00 70
Dinner 18:00 50 19:00 60 18:30 80
Snack 22:00 30 22:30 25 21:30 20

Table 15.2 Exercise scenario for one-hour duration for 3-day closed-loop
experiment using mGIPsimdthe integrated multivariable metabolic and
physiologic simulator.
Exercise First day Second day Third day
Morning Treadmill at 10:30 Bicycling at 09:45 Treadmill at 10:00
Afternoon Bicycling at 16:00 Treadmill at 16:45 Bicycling at 16:15