Anaesthetic Modelling And Closed-loop Strategy For Anaesthesia Depth Monitor

General anesthesia refers to the reversible functional inhibition when a proper amount of anesthetic drugs are applied to the central nervous system. Anesthesia creates a deep level of sedation where a patient is unresponsive to any stimulation including the pain associated with a surgical incision. The specific purposes of anesthetic drugs includes pain relief, blocking memory of the procedure, relaxing the muscles of the body and inhibiting normal body reflexes to make surgery safe and easier to perform.During the clinical surgery, anesthesiologists need to take advantage of the beneficial effects of different anesthetic drugs through the adjustment and control of dosage and timing to achieve a balanced anesthesia, and the assessment of conditions of the patient often needs to be made based on anesthesiologists’ clinical experience. From a clinical perspective, the methodology of traditional manual injection of anesthesia medications, which relies heavily on the personal experience of the anesthesiologists, is unquantifiable and lacks unified standards; therefore it is difficult to guarantee the effectiveness of anesthesia. In addition, the fatigue caused by long operation hours and continuous surgeries may increase the anesthesiologists’ frequency of errors and thus increase surgical risk.With the emergence of a variety of indexes to monitor the depth of anesthesia, people had started the research on anesthesia closed-loop control, with the purpose of realizing automated drug infusion and adjustment through feedback strategies. However, human drug metabolism is a complex nonlinear function and this has increased the technical difficulties of closed-loop control of anesthesia. This research is aimed to improve the stability of anesthesia, to shorten induction and recovery time, to lighten the workload of the anesthesiologist while improving efficiency and to reduce surgical risk. In this paper, in-depth research has been conducted on the population pharmacokinetic modeling methods of anesthetic medicine, anesthetic conventional quick cure propofol, drug model of remifentanil and anesthesia closed-loop control. Thus this paper is of both theoretical and clinical significance. Focusing on population pharmacokinetic models of propofol and remifentanil, this paper concentrates on the nonlinear mixed-effect modeling and the application of artificial neural network modeling, based on which the closed-loop fuzzy control of depth of anesthesia and nonlinear sliding mode control are built, using cerebral state index as feedback variables. Through theoretical analysis and simulation, research was carries out on the anesthesia closed-loop strategies. Main contributions and innovations of this paper are summarized below:(1) Summarized the population pharmacokinetic algorithm principles and elaborated on the nonlinear mixed-effect modeling methods commonly used by current anesthesia medication. By improving the SAEM-MCMC algorithm, two common clinical intravenous anesthesia drugs of the propofol and remifentanyl models are optimized. And the fixed effects and random effects of the propofol and remifentanil as well as models’selection principles are also analyzed.(2) Regarding the strong time-varying characteristics of anesthesia medication, as well as complex atrioventricular structures and low accuracy in the prediction of blood drug concentration, the artificial neural network prediction models for propofol and remifentanil blood concentrations are established. Based on factors affecting narcotic drugs’ blood drug concentration and the characteristics of its medicine pharmacokinetics models, systematic research was done on the construction of pharmacokinetics models in the artificial neural network for the prediction of blood drug concentration. The paper proposed drug concentration models based on improved Elman neural network and carried out multiple sets of verification using related data.(3) Built the anesthesia closed-loop fuzzy control system which uses cerebral state index as feedback variables, and proposed the fuzzy rules and shape of membership function between cerebral state index errors and output rates optimized by the particle swarm algorithm. Besides, a simulation study is conducted, which obtained positive control effects.(4) Regarding issues during the anesthesia processes including random frequent fluctuations and the surgical interventions appeared, the paper introduced nonlinear sliding mode control theory in the anesthesia close-loop fuzzy control system, to establish self-adaptive integral sliding mode closed-loop controller, which proved the validity of the research results by simulation, with good robustness.Through the four above-mentioned aspects of this study, the paper had shown that both the improved nonlinear mixed effect and neural network methods can achieve relatively high precision on anesthesia quick-effect drug modeling and blood drug concentration prediction. Additionally, the anesthesia close-loop fuzzy control and nonlinear sliding mode control system based on cerebral state index had indicated a broad application prospect in the simulation experiments.