Application of Model Predictive Control and Moving Horizon Estimation for the Development of Closed-Loop Inhaled Anesthetics Administration

This work presents the simultaneous application of the model predictive control (MPC) and the moving horizon estimation (MHE) for the development and simulation of a closed-loop control system that allows the administration of inhaled anesthetic (sevoflurane), considering the hypnotic depth, through the bispectral index (BIS), as a control variable. The controller and the state estimator implement a PK-PD model that represents the dynamics of the system. The control system is tested in a set of 11 virtual patients generated from a PBPK-PD model and individualized in accordance with demographic characteristics and variations in the values of the PD model parameters. The control strategy developed is characterized by a rapid response to disturbances and a short induction time without overshooting, an appropriate dosage and robustness to bring and maintain the BIS at the reference value. The above is evaluated by means of various quantitative metrics.