Machine learning for a combined electroencephalographic anesthesia index to detect awareness under anesthesia

Abstract

Spontaneous electroencephalogram (EEG) and auditory evoked potentials (AEP) have been suggested to monitor the level of consciousness during anesthesia. As both signals reflect different neuronal pathways, a combination of parameters from both signals may provide broader information about the brain status during anesthesia. Appropriate parameter selection and combination to a single index is crucial to take advantage of this potential. The field of machine learning offers algorithms for both parameter selection and combination. In this study, several established machine learning approaches including a method for the selection of suitable signal parameters and classification algorithms are applied to construct an index which predicts responsiveness in anesthetized patients. The present analysis considers several classification algorithms, among those support vector machines, artificial neural networks and Bayesian learning algorithms. On the basis of data from the transition between consciousness and unconsciousness, a combination of EEG and AEP signal parameters developed with automated methods provides a maximum prediction probability of 0.935, which is higher than 0.916 (for EEG parameters) and 0.880 (for AEP parameters) using a cross-validation approach. This suggests that machine learning techniques can successfully be applied to develop an improved combined EEG and AEP parameter to separate consciousness from unconsciousness.

Fig 1. Schematic overview of the study protocol.

LOC: Loss of consciousness; ROC: Return of consciousness. Time frame duration: 10s for EEG parameters, variable for AEP parameters. Duration between two requests to push the hand: 30s. The actual surgery period is not depicted in this figure, it took place between LOC₂ and ROC₂.

Fig 2. High-dimensionality performance.

Performance comparison on large parameter sets between a SVM and a Naive-Bayes Classifier. The SVM is able to tolerate high-dimensional data input, while the performance of the Naive-Bayes-Classifier is decreasing.

Fig 3. Pure EEG vs pure AEP indicators.

Comparison of the prediction performances using the RBF-Kernel SVM reached by allowing only either EEG or AEP parameters, or combinations of both. The single EEG and AEP parameters with the highest P_K are shown for comparison.

Fig 4. Influence of the EEG low pass filter setting on prediction probability.

Comparison of the influence of the EEG low pass filter settings (fhigh) on prediction performance. The figure shows that an increase in the EEG signal range leads to an increase in prediction accuracy. The single EEG parameter PeEn (permutation entropy) is shown for comparison.