Prediction of outcome in traumatic brain injury patients using long-term qEEG features

Treatment of patients suffering from severe traumatic brain injury (TBI) commonly involves sedation and mechanical ventilation during prolonged stay in the intensive care unit. Continuous EEG is often monitored in these patients to detect epileptic seizures. It has also been suggested that EEG has prognostic value regarding the outcome of the treatment. In this study the ability of 186 qEEG features to predict the outcome of the treatment of TBI patients is assessed. The features are based on the power spectrum of the EEG. The data underlying the study contains long term (over 24 h) recordings from 20 patients treated in the postoperative intensive care unit of the North Estonian Medical Center. 12 qEEG features were found to have predictive value when evaluated by calculating the area under the receiver operating curve constructed from feature probabilities.