LAUSR

Abstract Schizophrenia and autism share some genotipic and phenotypic aspects as connectome miswiring and common cognitive deficits. Currently, there are no medical tests available for either disorders, and diagnostics for both of them include direct reports of relatives and clinical evaluation by a psychiatrist. Despite several medical imaging biomarkers have been proposed in the past, novel effective biomarkers or improvements of the existing ones is still need. This work proposes a dynamic functional connectome analysis combined with machine learning techniques to complement the present diagnostic procedure. We used the moving window technique to locate a set of dynamic functional connectivity states, and then use them as features to classify subjects as autism/schizophrenia or control. Moreover, by using dynamic functional connectivity measures we investigate the question whether those two disorders overlap, namely whether schizophrenia is part of the autism spectrum and which brain region could be involved in both disorders. The results reveal that both static and dynamic functional connectivity can be used to classify subjects with schizophrenia or autism. Lastly, some brain regions show similar functional flexibility in both autism and schizophrenia cohorts giving further possible proofs of their overlaps.