Passivity based control of blood glucose: an alternative closed-loop strategy for the treatment of type 1 diabetes
Résumé
Two main types of control algorithms have been evaluated to determine insulin delivery in a closed-loop system for the treatment of type 1 diabetes: the Proportional-Integral-Derivative control (PID) attempts to emulate beta-cell function whereas the Model Predictive Control (MPC) is based on a dynamical model of the glucose-insulin interactions. The aim of our study is to evaluate Passivity Based Control (PBC) as an alternative closed-loop strategy. PBC uses the energy principle of damping addition to ensure a robust stability of closed-loop systems despite parametric uncertainties and disturbances. It benefits from intuitive tuning criteria and provides explicit state feedback control laws which require much less computational resources than MPC, since they need no optimisation algorithm. This preliminary work proposes the design of a passivity based controller which can be calculated either for the intravenous or the subcutaneous route, using a modified version of Bergman model. To date we have tested this new strategy in simulation with Matlab/Simulink, over one day including three typical meals as disturbances. Its early theoretical results are encouraging, showing the controller ability to maintain the glucose values at the target level while keeping realistic basal and bolus insulin delivery. Of course, further studies are required to validate the approach. Especially, it will be necessary to examine the practical feasibility of a state feedback control knowing that only the glucose measurement is available