On an Automatic Intravenous Anesthesia Control System

近年来，开发出一种副作用小、特性好的静脉麻醉药异丙酚。在本研究中，我们研究了麻醉深度的指标和使用异丙酚的静脉麻醉控制方法，以开发一种自动静脉麻醉控制系统，以使患者的麻醉深度保持在一个适当的水平，以最少的药物输注。具体地说，我们研究了1）麻醉深度的适当指标，2）麻醉深度对异丙酚输注的响应的数学模型，3）采用模型预测控制的麻醉深度控制系统。结果表明：脑电双频指数（BIS）是心血管动力学指标、中位功率频率（MF）、频谱边缘频率（SEF95）和BIS中最好的麻醉深度指标，根据麻醉药物输注量和BIS的数据，建立了麻醉药物输注量、血药浓度、该模型由带时滞的四阶传递函数和非线性函数组成。此外，由于死区时间和描述效应室浓度与BIS之间关系的非线性函数具有很大的个体差异，我们开发了一种从麻醉导入过程中麻醉药物输注和BIS的数据中识别它们的方法。我们开发了一个基于模型预测控制的静脉麻醉控制系统。为了评估我们的系统的准确性和可靠性，我们在与实际操作类似的干扰下进行了模拟。实验结果表明，该系统具有良好的性能。

Recently, an intravenous anesthetic drug, propofol, which has less side effects and better characteristics than the other anesthetic drugs is developed. In this research, we have studied an index of anesthetic depth and an intravenous anesthesia control method using propofol in order to develop an automatic intravenous anesthesia control system for keeping patients' anesthetic depth at an appropriate level with minimal infusion of drug. To be concrete, we have studied 1) an appropriate index of anesthetic deptlt, 2) mathematical model of the response of the anesthetic depth to propofol infusion, and 3) a control system of the anesthetic depth using model predictive control. The obtained results are as follows.Bispectral index (BIS) is the best index of anesthetic depth among the index of cardiovascular dynamics, median power frequency (MF), spectral edge frequency (SEF 95) and BIS.From the data of anesthetic drug infusion and BIS during surgical operations, we made a mathematical model which describes the relation among the anesthetic drug infusion, the drug concentration in blood, effect-site concentration and BIS.The model consists of four-order transfer function with dead time and nonlinear function. Furthermore, since the dead time and the nonlinear function which describes the relation between effect-site concentration and BIS have much individual difference, we have developed a method of identifying them from the data of anesthetic drug infusion and BIS during anesthesia introduction.We have developed an intravenous anesthesia control system using model predictive control. To evaluate the accuracy and reliability of our system, we made simulations under similar disturbances as in the actual operations. The results indicated that our system has enough performance.