Neural network method to determine the vigilance levels of the central nervous system, related to occupational chronic chemical stress

The effects of chronic toxic occupational factors and functional disorders of the central nervous system (CNS) in chemical industry were studied. These factors cause various stages of chronic chemical stress on the human CNS together with changes of the vigilance levels. On the basis of QEEG data analysis and psychometric tests we identified three stages of occupational chemical stress syndromes according to the CNS vigilance level (ordered from light to severe): hypersthenic syndrome, hyposthenic syndrome, and organic psychosyndrome. Each syndrome is characterized by specific changes in the QEEG data. A perceptron-based neural network was developed for the classification of the QEEG data to one of the above-mentioned syndrome classes. The data of 77 patients and 10 healthy subjects were selected to test the algorithm. Different combinations of the QEEG data as input features to the classifier were chosen. The most reliable classification was obtained when QEEG data measured during the visual stimulation of the CNS were used. However, sometimes the algorithm was unable to solve the classification problem, or it took a very long time to train the perceptron. In part, difficulties arose from using a perceptron-based algorithm, which can classify only linearly separable data.