Performance analysis of neural network based control of hypnosis and analgesia during anesthesia by employing a pharmacokinetic- pharmacodynamic model

Monitoring and controlling the hypnosis and arterial pressure during a surgery is really vital since in excess of dosing and below dosing can be hazardous for the patients. Anesthesia drugs have impact on multiple results of an anesthesia patient. Automation of anesthesia is very useful as it will provide more time and flexibility to anesthesiologists to focus on critical issues that may arise during the surgery. Furthermore patient safety and cost reduction. Anesthetics are administered to regulate hypnosis and analgesia, respectively in the patient during the surgery. Most distinctive measures include Bispectral index (BIS), mean arterial pressure (MAP) and in general, BIS and MAP as the indirect measurements of hypnosis and analgesia, respectively. Isoflurane is given as the input to the Pharmocokinetic-pharmacodynamic model (PK-PD), from the model BIS and MAP were taken as output. In this work, a neural network based internal model controller (NN-IMC) is proposed by regulating the level of hypnosis and pressure. Performance of proposed approach is evaluated with conventional Proportional-Integral (PI) controller. Simulation results show that proposed NN-IMC outperforms conventional PI controller.