Schizotypy refers to a set of temporally stable traits that are observed in the general population and that resemble, in attenuated form, the symptoms of schizophrenia. In a previous work, we identified volumetric patterns in thalamic subregions which were associated with disease status, and trained a random forests classifier, accounting for such thalamic volumetric patterns, that discriminated healthy controls (HC) from patients with schizophrenia (SCZ) (81% accuracy) [1].

i) to assess performance of random forests classifier developed by Pergola and coworkers [1], in an independent sample of healthy subjects; ii) to test whether false positives (FP), i.e. HC classified as SCZ based on such classifier would be associated with greater schizotypy compared with true negatives (TN), i.e. HC classified as such.

A total of 167 HC participated in the MRI study and filled the Schizotypal Personality Questionnaire (SPQ). We pre-processed MRI data with SPM8 and DARTEL. Then, we used thalamic grey matter volumes (GMV) as features in the random forests prediction of disease status at the single subject level. Finally, we tested SPQ scores differences between FP and TN with Mann-Whitney test.

The classification accuracy was 71%. FP had greater SPQ scores compared to TN (P = 0.007).

Classification accuracy of our classifier in an independent sample suggests that thalamic GMV patterns are reproducible markers of disease status. Furthermore, the present results also suggest that variability of thalamic GMV patterns in HC may have relevance for subclinical phenotypes related to schizophrenia spectrum.

The authors have not supplied their declaration of competing interest.