Estimating Correlation Between Brain Consciousness and Depth of Anesthesia Based on EEG

Electroencephalography (EEG) is an intuitive indicator forjudging the patient's level of consciousness and the depth of anesthesia (DOA) in clinical surgery. Evaluating the patient's DOA according to EEG can assist the anesthesiologist to more accurately determine the type and dose of anesthetics. However, due to the irregularity of EEG waveform, it is difficult to directly give the correlation between the patient's level of consciousness and DOA. From this point of view, in this article, the nonlinear dynamic methods and the frequency-domain methods are combined to analyze EEG, and characteristic parameters are used to analyze the changing trend of the patient's level of consciousness during general anesthesia. Then the Elman Neural Network (ElmanNN) and the Long and Short-Term Memory Neural Network (LSTM) are used to extract the temporal features of the characteristic sequence segments and identify the patient's DOA in the current period of time and predict the DOA in the next period of time. The results show that in the recognition and prediction of the DOA, when the aforementioned characteristic indicators are input as the combined parameter, the test accuracy of the ElmanNN is 88.46% and 82.69%, and the test accuracy of the LSTM is 90.38% and 74.04%. Compared with other feature combination methods, this model is more accurate in determining the patient's DOA.