This paper introduces the use of cross recurrence rate (CRR) for discriminating between ‘conscious’ and ‘unconscious’ state from the electrical brain activity (EEG) of patients under general anesthesia. CRR measures the bivariate dependencies between two time series by looking at recurrences of the phase space from one system to the other. Here, the feasibility of pairwise electrode CRR to discriminate ‘conscious’ and ‘unconscious’ state is investigated using the EEG of 10 subjects during recovery of consciousness from general anesthesia with propofol. Performance is investigated with a Support Vector Machine. An average accuracy of 98% is obtained.
