An augmented,
adaptive algorithm utilizing
model predictive control (MPC) is developed for closed-loop
glucose control in type 1 diabetes. A linear empirical input-output subject model is used with an MPC
algorithm to regulate blood glucose online, where the subject model is recursively adapted, and the
control signal for delivery of
insulin and a counter-regulatory agent such as
glucagon is based solely on online glucose concentration measurements. The MPC
signal is synthesized by optimizing an augmented objective function that minimizes local
insulin accumulation in the subcutaneous depot and
control signal aggressiveness, while simultaneously regulating glucose concentration to a preset reference
set point. The mathematical formulation governing the subcutaneous accumulation of administered
insulin is derived based on nominal temporal values pertaining to the
pharmacokinetics (time-course of activity) of insulin in human, in terms of its
absorption rate, peak
absorption time, and overall time of action. The MPC
algorithm is also formulated to provide control action with an integral effect, and in essence minimizes overall
drug consumption. When employed as a modulator in an automated integrated glucose-
control system for type 1 diabetes, the
control algorithm provides the
system with self-learning capability that enables it to operate under unrestricted activity of the subject.