We studied the complexity of the electroencephalogram (EEG) in schizophreni
c patients by estimating the first Lyapunov exponent (L1), which might serv
e as an indicator of the specific brain function in schizophrenia. We recor
ded the EEG from 25 schizophrenic patients (12 male, 13 female, age = 25.1
+/- 7.0 years) fulfilling DSM-IV criteria and 15 healthy controls (9 male,
6 female; age = 27.8 +/- 4.2 years) at 16 electrodes, different from previo
us studies which recorded the EEGs at limited electrodes. We employed a met
hod with an optimal embedding dimension to calculate the L1s. For limited n
oisy data, this algorithm was strikingly faster and more accurate than prev
ious ones. Our results showed that the schizophrenic patients had lower val
ues of the L1 at the left inferior frontal and anterior temporal regions co
mpared with normal controls. These results for L1 in non-linear analysis ha
ve some differences from those for power ratios in linear analysis. These s
uggest that the non-linear analysis of the EEGs such as the estimation of t
he L1 might be a useful tool in analyzing EEG data to explore the neurodyna
mics of the brains of schizophrenic patients. (C) 3000 Elsevier Science Ire
land Ltd. All rights reserved.