Offset-free hybrid model predictive control of bispectral index in anesthesia

Abstract

This paper deals with the anesthesia control, using bispectral index (BIS) as a measure of the depth of anesthesia; controlled by hybrid model predictive control strategy. The piecewise affine (PWA) hybrid pharmacodynamic model of a patient containing a set of local linear dynamics is used to describe the relationship between BIS value and drug infusion rate. The hybrid model predictive control problem is formulated as a mixed integer quadratic programming (MIQP) problem and solved online. Furthermore, a disturbance observer is designed for the offset-free BIS reference tracking. The results of designed controller intended for reference tracking, disturbance rejection and constraint handling are presented. Moreover, the performance of nonlinear MPC is compared with the hybrid MPC and is shown to be computationally less complex and fast.