Abstract
With a rise in the number of patients being diagnosed with Type 1 Diabetes, there is a pressing need to develop insulin therapies to treat this chronic disease. Studies have shown that failure to maintain blood glucose levels can have long term negative consequences on a person's health. Extensive research aimed at designing controllers and optimizing insulin delivery methods has been conducted. With the Food and Drug Administration (FDA) approving the use of simulation models for in-silico trials, efforts are being made to develop models that can describe the insulin-glucose dynamics effectively. The majority of Type I Diabetes patients rely on insulin pumps that have proven to be very efficient in the treatment of diabetes. However, these devices still involve some degree of human intervention making them prone to errors.This research proposes the use of Particle Swarm Optimization (PSO) algorithm to optimize the basal insulin delivery and a PID controller to maintain glucose levels within acceptable limits for patients suffering from Type 1 Diabetes. The goal of the algorithm and the controller proposed in this research is to mimic the function of a pancreas in achieving ideal glucose profile The PSO algorithm is designed to account for variations in the physiological parameters of patients. The PID controlled closed loop system is subjected to a series of robustness tests to evaluate its effectiveness in the face of external disturbances. The gains of the PID controller are tuned using the optimization algorithm and its performance is compared with that of a manually tuned controller. Simulation results show that the PSO algorithm is effective in optimizing insulin delivery and a properly tuned PID controller can be used to correct variations of glucose profile resulting from external disturbances. It is suggested that a PSO optimized basal insulin delivery profile coupled with feedback controlled corrective bolus input can result ineffective treatment of Type 1 Diabetes.