Patient-specific MPC for improved robustness in anesthesia

This paper presents a Model Predictive Control (MPC) approach for Depth-of-Hypnosis (DoH) control in Total IntraVeneous Anesthesia (TIVA), where the Bispectral Index (BIS) signal is the process variable. In particular, a Branch and Bound (BnB) algorithm is employed after the induction phase for the identification of the parameters of the propofol pharmacokinetic/pharmacodynamic (PK/PD) model for the patient. Then, a suitably designed predictive controller based on that model is applied during the maintenance phase so that the robustness to inter-patient variability is properly addressed. Extensive simulation results show the effectiveness of the individualized approach.