Vigilance monitoring from the EEG power spectrum with a neural network

Length of vigilance is a major obstacle in jobs associated with low levels of arousal. To provide the highest levels of safety, the level of attention, particularly visual awareness, has to be monitored. The infrequent, but catastrophic disasters caused by aviation controller mistakes could be avoided with attention monitoring. Furthermore, the recent explosion in the diagnosis of Attention Deficit Disorder and related learning disabilities has increased the demand for an attention monitoring tool, possibly as a screening tool or a therapeutic biofeedback mechanism. A system has been designed, off-line, as a precedent to a real-time awareness predictor. The electroencephalograph (EEG) is used as the major predictive data with a radial basis function network classifying the attention level. Two different receptive field placements were used for each subject. Using a system trained to each subject, the results show the ability to predict performance from the EEG spectrum. The subject´s habituation was directly correlated to the accuracy of the system