Frontal dysfunction in schizophrenia - a new electrophysiological classifier for research and clinical applications

We determined whether schizophrenic patients can be reliably classified wit h electrophysiological tools. We developed a fully computerized classifier based on 5 minutes of EEG recording during an acoustical choice reaction ti me task (AMDP-module IV). We included factorized variables from the frequen cy domain and evoked potentials (N100/P200-complex) from central and fronta l electrodes, which were preprocessed in a sample of 150 normal subjects pr ior to classification. We applied discriminant analyses to the electrophysi ological data from depressive, schizophrenic and schizotypal subjects, most of them being unmedicated or drug-naive. The classifier was developed on a training set (33 schizophrenics, 49 normals) and tested on an independent sample (32 schizophrenics, 49 normals). A simple three-variable classifier was found to classify schizophrenics and normals in 77 % of those tested co rrectly. Diagnostic specificity of the das sifier proved to be low as the i nclusion of depressive patients (n = 60) significantly decreased classifica tion power. It was demonstrated that atypical but not typical neuroleptic d rugs may influence the classification results. Correctly classified schizop hrenics showed significantly more negative symptoms and slower reaction tim es than those schizophrenics who were misclassified as normals. In contrast , these misclassified schizophrenics showed a non-significant trend for mor e positive symptoms and shorter reaction times. As the correctly classified schizophrenics showed increased frontally pronounced delta-activity and de creased signal power of the N100/P200 amplitude, it was concluded that thes e schizophrenics show dysfunction of the frontal lobe. It is proposed that this new classifier can be useful for clinical and research applications wh en subtyping of schizophrenics with detection of frontal dysfunction as the aim.