Toward a model-free feedback control of blood glucose induced by meals for type-1 diabetics: in silico studies

This work addresses the problem of developing control algorithms for type 1 diabetic patients that provide an automatic connection between continuous glucose monitoring and insulin injection, without any manual intervention by the patient. This emerging technology is known as artificial pancreas (AP). In this study, a recent Model-Free Control (MFC) approach and its related “intelligent” regulators were exploited for this biomedical issue using an Extended Bergman Minimal Model. MFC is used to regulate the blood glucose (BG) level as well as to reduce postprandial glucose peak induced by meals in subjects with type 1 diabetes. Effects of unannounced meals three times a day are taken into account and are investigated. In silico results globally illustrate the feasibility of such a control. The proposed controller has proven its effectiveness in reducing postprandial BG rise, eliminating the effects of external disturbances such as the effect of meals, and maintaining a healthy BG level under uncertainties.