Analysis Of Multiple Physiological And Biochemical Information And Study Of Automatic Classification In Schizophrenia Patients

Schizophrenia(SZ)has characteristics of complex etiology,high prevalence and high disability rate.It is of great significance to explore the pathogenesis and neuropathological mechanism of SZ for early diagnosis,early intervention and early treatment.In this study,the physiological and biochemical information,such as EEG,intestinal microbiota and blood,of schizophrenia patients(SZs)and normal controls(NCs)were acquired and analyzed.Thus,the between-groups differences,correlations and mediation effects among the properties of functional brain network(FBN),the abundances of intestinal microbiota,and the indicators of immunity and oxidative stress were analyzed.Combined with all the above characteristics,the automatic classifications between SZs and NCs were performed using five machine learning algorithms and the features with high weight were analyzed and discussed.Forty-nine SZs and fifty NCs were included in this study,in which the samples of EEG,feces and blood were collected.The properties of FBN were analyzed based on phase-locked value and graph theory.The abundances of intestinal microbiota were analyzed based on 16 S r RNA sequencing.The counts of blood cell and the indexes of oxidative stress were obtained based on biochemical analyses.First,the results of between-groups differences were as follows.1)The FBN of SZs showed abnormalities of small world attributes in the frequencies,such as ?1,?2,?.Compared with NCs,SZs showed significant low values of the nodal centralities in the prefrontal,frontal,and parietal lobes,which implied the topological organization of FBN in SZs was changed and the efficiency of information transmission in the above brain areas was reduced.2)There were 18 species of bacteria with significant between-groups difference in the abundance of genera level,which suggested an imbalance of intestinal microecology in SZs.There was a significant positive correlation between the abundance of ruminococcus and the positive scores of PANSS,which might be related to the influence of ruminococcus on the secretion of serotonin.3)The values of monocyte-lymphocyte(MLR),superoxide dismutase and C-reactive protein were significantly increased in SZs,indicating that the immune system and oxidative stress system of SZs might be in an activated status for chronically.Second,the correlations among the physiological and biochemical information were also analyzed.The abundances of eubacterium,ruminococcus and mogibacterium were related with the shortest path length,local efficiency,and global efficiency of FBN.The inflammatory indexes of PLR,the oxidative stress of SOD and CPR were positively correlated with the shortest path length of FBN and negatively correlated with the global efficiency of FBN.The abundance of ruminococcus and mycobacterium were related with white blood cells,neutrophils,SOD and CRP.Third,the analysis of mediating effect demonstrated that the abundance of ruminococcus fully mediated the global efficiency of FBN through white blood cells,and partially mediated the clustering coefficient,local efficiency,and global efficiency of FBN through SOD.These results suggested that the effects of intestinal microbes on the function of human brain might be mediated by chemical substances.Automatic classifications of SZs and NCs were studied using five machine learning algorithms,in which the characteristics of three physiological and biochemical information and their combinations were used as input features.The performance of different algorithms and feature combinations were also compared.The method of k-nearest neighbor showed the best performance using the combination of all information as input features,in which the classification accuracy of an independent test including 24 subjects was 100%.Moreover,the weights in the classification of SOD was the highest,followed by the blood index of CRP and MLR,as well as the nodal centralities of FBN in the temporal lobe and several features of the abundance of intestinal microbiota.The results of this study suggest that intestinal microbiota may affect the function of human brain by interfering with immune system and oxidative stress system.Analyses of physiological and biochemical information are useful to the exploration of the pathogenesis and neuropathological mechanism of SZ and the development of objective evaluation indicators for the diagnosis of SZ.