A New Model Predictive Control for the Artificial Pancreas

Closed-loop insulin delivery system have led to significant improvement in the quality of life of subject with diabetes and are challenging to overcome the barrier of hypoglycemia, the most frequent complication of insulin therapy. The reliability of the system, composed by a computer algorithm, a glucose sensor and an insulin infusion device, depends on the knowledge and predictor capacity of the physiology of blood glucose regulation. This paper describes the physical-mathematical fundamentals and the most important results of a new three-compartmental model. The model includes exogenous insulin injected in subcutaneous tissue with local degradation, three explicit delays and three influencing physiologically-based parameters controlling the regulatory system. The parameters have been calculated through the simulation of actual clinical data and, therefore, implemented into the mathematical model to successfully simulate the clinical data obtained at Campus Biomedico in the normal-life regulation (1 day and 4 days) of diabetic patients. The estimated model parameters were physiological meaningful and provided insights on the subject's dysfunction.