Personalized asymmetric multiple PID to automatize the procedure of intravenous general anesthesia

Abstract

General anesthesia has improved considerably over the years due to the automation of drug perfusion through closed-loop control which provided numerous advantages regarding efficiency and patient safety. This work describes the implementation and Validation of a control algorithm based on a multiple-input and multiple-output proportional–integral–derivative (MIMO-PID) controller. The main objective is to keep patients closer to the desired range of monitor values into satisfactory anesthetic state.The MIMO-PID decides the drug perfusion of propofol, remifentanil, and rocuronium based on the bispectral index, noxious stimulation response, and neuromuscular blockade monitorization. The main MIMO-PID variant is based on a feed-forward target which is used to define the asymmetric errors. The asymmetry allows different errors to be defined in each branch of each single-input and single-output proportional–integral–derivative. The MIMO-PID is tuned using the criterion of integral absolute error and their performance is evaluated through a range of experiments applying different metrics to evaluate monitors and drug perfusions. The satisfactory anesthetic state is achieved during the induction phase of anesthesia; however, we also had situations with a low relative deviation from the registered median values. We obtained good results by using a strict disturbance rejection.The controller was able to mitigate the surgery influence on the anesthetics effects when there were less than 20, the patient remained close to the desired anesthetic state. In the presence of significant or more numerous stimuli, the controller was able to achieve a satisfactory anesthetic state within a short period of typically less than 7 min.