A simulation study of an adaptive inverse controller for closed-loop control in type 1 diabetes

Closed-loop control algorithms in diabetes aim to calculate the optimum insulin delivery to maintain the patient in a normoglycaemic state, taking the blood glucose level as the algorithm's main input. The major difficulties facing these algorithms when applied subcutaneously are insulin absorption time and delays in measurement of subcutaneous glucose with respect to the blood concentration. Additionally, closed-loop algorithms should be able to adapt to interand intra-patient deviations. This paper presents and adaptive closed loop algorithm (ACIC) and compares its performance with the static version of the same algorithm (CIC) obtained by inversion of an existing mathematical model. The simulation studies use an in silico population simulated with a different model. The results demonstrate that the adaptation strategy reduces the glucose variability. The presented adaptation strategy should be considered as an adequate mechanism for any other metabolic control algorithm.