A study of schizophrenia inheritance through pattern classification

In the present study, functional connectivities throughout the whole brain were examined in schizophrenic patients (Sch), their healthy siblings (HS) and healthy controls (HC) using resting-state functional magnetic resonance imaging. One-against-one classifications using PCA + nonlinear SVM were then employed between the three groups using the functional connections as original features, resulting in 78.26% classification accuracy between HC and Sch, 73.47% classification accuracy between Sch and HS, and 63.83% between HC and HS, which were proven effective through 10000 times permutation test. The results have given the evidence that the healthy siblings of schizophrenic patients showed higher risk for developing schizophrenia than healthy controls to some extent from the aspect of classification, reflecting the inheritance of schizophrenia.