BP网络在麻醉深度监测算法上的应用

麻醉是临床手术中必不可少的环节，但麻醉的过深或过浅可能给病人带来伤害，因而对麻醉深度的监测具有较高的临床价值。脑电是目前检测麻醉深度最有潜力的方法，首先通过滤波等处理方式得到较为纯净的脑电信号，分析时域和频域的特征，计算相应的参数，并将该参数作为前向反馈神经网络的输入参数，选择合适的BP神经网络拟合得到一个能够评价麻醉深度的无量纲常数。使用BP神经网络拟合结果来表征麻醉深度准确率普遍在90%以上，反映了BP神经网络在麻醉深度监测上具有较高的应用价值。

Application of back-propagation network in algorithm for monitoring depth of anesthesia

Abstract: Anesthesia is an essential part of clinical operation, but the inappropriate depth of anesthesia (overdose or underdose) may cause harms to patients. Therefore, monitoring the depth of anesthesia has a high clinical value. Currently, electroencephalogram is the most potential method to detect the depth of anesthesia. After obtaining the pure electroencephalogram signal by filtering and other processing methods, the time-domain and frequency-domain characteristics are analyzed, and the corresponding parameters are calculated. Then the parameters are used as the input parameters of feedforward neural network, and a dimensionless constant which can evaluate the depth of anesthesia is obtained by back-propagation (BP) neural network fitting. The accuracy of using BP neural network fitting results to characterize the depth of anesthesia is generally more than 90%, which reflects that BP neural network has a high value in monitoring the depth of anesthesia.