Characteristics Of EEG Signals Under The Cognitive Load Of Schizophrenic Patients

Cognitive impairment is one of the three primary symptoms of schizophrenic patients.It also shows important value in early diagnosis and warning for high-risk individuals.Moreover,the improvement of cognitive impairment is one of the most important characters to estimate efficacy of schizophrenia and its long-term prognosis.However,the existing methods of cognitive impairment assessment are dominated by patient’s complaint and scale assessment,which are strongly subjective and one-sided,in order to overcome these deficiencies,in this study,we creatively designed a set of experimental procedures under cognitive load based on a classical oddball paradigm using numbers instead of traditional tone.We also investigated the characteristics of EEG signals in schizophrenic patients under the cognitive load by using nonlinear dynamic analysis,functional network analysis,and machine learning analysis methods,aiming at finding characteristics of EEG signals which could reflect the improvement of cognitive impairment in schizophrenic patients.There were 17 schizophrenic patients and 19 healthy persons participated in this experiment.For nonlinear dynamic analysis,by comparing the differences of correlation dimension and sample entropy between patients group and healthy group,we found that: under the cognitive load,the values of correlation dimension and sample entropy of patients were lower than those of healthy subjects;significant differences between two groups were primarily in the prefrontal cortex,especially in the γ rhythm;abnormal function of patients were found in the left frontal brain region;in the β rhythm,sample entropy values of patients in all leads were higher than those of healthy subjects;in the θ rhythm,significant differences were found in the prefrontal cortex and occipital lobes between groups;in the α rhythm,the outcomes of the temporal lobe were significantly higher than those of other brain regions,while no significant difference between groups was found.For functional network analysis,by comparing the differences of global and node properties between patients group and healthy group,we found that: compared with healthy subjects,the patients showed disrupted small-world network properties under the cognitive load;particularly,the patients showed decreased clustering coefficient,increased shortest path length,decreased global and local efficiencies.The results of node properties showed that the frontal and temporal lobes were highly related to cognitive impairment in schizophrenia patients.Main findings of the automatic classification analysis based on support vector machine included: in order to distinguish the patients group from the healthy group,the properties of functional networks and the characteristics of nonlinear dynamic were used as the input values of classifications,thus the best performance was achieved.Compared with the nonlinear dynamic characteristics,the properties of functional networks resulted in a better performance of classification.Our results indicated that the functional disruptions in the frontal lobe might be the important factors of cognitive impairments in schizophrenic patients.Our study may provide more convenient and economical parameters to assess cognitive functions,which can be developed into a new biomarker for early screening and diagnosis of schizophrenia.