Bispectral Analysis of Electroencephalogram Using Neural Networks to Assess the Depth of Anesthesia

Artificial neural networks (ANNs) are used to perform bispectral analysis and form a conclusion on the degree of patient brain activity. The article also shows individual results of functioning of the algorithms on real EEG signals and compares these results with expert judgments of doctors (anesthesiologists and neurophysiologists).

The efficiency of current sedativeassisted treatment methods essentially depends on the optimal dosage of these drugs. As a rule, the optimal dose means the minimum dose that ensures the safety and efficacy of treatment. The requirement of the minimum dose is usually due to two main points. On the one hand, it ensures quicker emergence of a patient from anesthesia while minimizing complications, and on the other hand, it saves expensive drugs. The optimal dose can be estimated in several ways, for example, by calculation taking into account the analysis of the patient’s hemodynamics (heart rate, blood pressure, oxygen consumption, etc.). A wellknown drawback of this approach is the fundamentally ambiguous relationship between the hemodynamic parameters of the patient and the depth of anesthesia.