Anwendung der rough set theorie zur unterscheidung von wach-und narkose-EEG

EEG derived parameters arc increasingly applied to detect intraoperative awareness during general anaesthesia. Since the ability of a single parameter to discriminate between "awareness" and "unconsciousness" is limited, combination of several parameters is necessary. Rough Sot Theory provides suitable methods for the reduction of a set of available parameters to parameters, that are necessary for the discrimination of the states "awareness" and "unconsciousness". In this process, rules are generated that determine these states based on parameter values. These methods are applied to data from a study in 40 patients, who underwent surgery under general anaesthesia. 52 parameters are calculated from EEG-signal segments of the states "awareness" and "unconsciousness". The Rough Set methods substantially reduce the number of parameters. The generated rules yield satisfying classification rates of about 85%.