Application of Back Propagation Neural Network and Information Entropy in Deep Detection of Anesthesia

Abstract

In order to study the monitoring of anesthesia depth during general anesthesia, the EEG (electroencephalogram) signals of 30 patients with laparoscopic general anesthesia were taken as the research objects. The approximate entropy, sample entropy, ranking entropy, and wavelet entropy of EEG signals under different anesthesia conditions were compared by BP (Back Propagation) neural network. The results showed that with the deepening of anesthesia, the four kinds of information entropies of EEG signal showed a downward trend. Among them, the sample entropy algorithm, ranking entropy algorithm, and wavelet entropy algorithm had a higher accuracy in the classification of anesthesia depth. Whereas, the network model established by combining sample entropy index and wavelet entropy index had the highest accuracy in judging anesthesia depth, which was 99.98%. To sum up, the method presented to monitor the depth of anesthesia by combining the characteristics of various EEG signals provides a new reference for the monitoring of the depth of anesthesia.