Machine learning and graph-based connectivity metrics identify language disruption in psychosis: Novel insights from dream reports in an Italian cohort
Schizophrenia Research October 20, 2025 Elena Camon, Stefano Masier, Natália Bezerra Mota et al.
Language disorganization in psychosis can be objectively measured by analyzing speech as a network. This study examined written and oral dream reports from 193 Italian participants (115 with acute psychosis and 78 healthy controls), constructing structural speech graphs and measuring connectivity attributes like the Largest Connected Component and its z-score relative to random graphs. Patients with psychosis showed significantly lower connectivity values than controls, with their speech graphs more closely resembling random word sequences. These differences remained after controlling for education and survived statistical correction. A Naive Bayes classifier using these features achieved 68% accuracy (AUC = 0.75), supporting automated detection. This work reinforces the cross-linguistic validity of graph-based analysis of thought and language impairments in psychosis.