Brain activity changes with emotional words in different stages of psychosis.

BACKGROUND: To date, a large number of functional magnetic resonance imaging (fMRI) studies have been conducted on psychosis. However, little is known about changes in brain functioning in psychotic patients using an emotional auditory paradigm at different stages of the disease. Such knowledge is important for advancing our understanding of the disorder and thus creating more targeted interventions. This study aimed to investigate whether individuals with first-episode psychosis (FEP) and chronic schizophrenia show abnormal brain responses to emotional auditory processing and to compare the responses between FEP and chronic schizophrenia. METHODS: Patients with FEP (n = 31) or chronic schizophrenia (n = 23) and healthy controls (HCs, n = 31) underwent an fMRI scan while presented with both emotional and nonemotional words. RESULTS: Using HC as a reference, patients with FEP showed decreased right temporal activation, while patients with chronic schizophrenia showed increased bilateral temporal activation. When comparing the patient groups, individuals with FEP showed lower frontal lobe activation. CONCLUSION: To the best of our knowledge, this is the first study with an emotional auditory paradigm used in psychotic patients at different stages of the disease. Our results suggested that the temporal lobe might be a key issue in the physiopathology of psychosis, although abnormal activation could also be derived from a connectivity problem. There is lower activation in the early stage and evolution to greater activation when patients become chronic. This study highlights the relevance of using emotional paradigms to better understand brain activation at different stages of psychosis.

Biased accuracy in multisite machine-learning studies due to incomplete removal of the effects of the site.

Brain MRI researchers conducting multisite studies, such as within the ENIGMA Consortium, are very aware of the importance of controlling the effects of the site (EoS) in the statistical analysis. Conversely, authors of the novel machine-learning MRI studies may remove the EoS when training the machine-learning models but not control them when estimating the models' accuracy, potentially leading to severely biased estimates. We show examples from a toy simulation study and real MRI data in which we remove the EoS from both the "training set" and the "test set" during the training and application of the model. However, the accuracy is still inflated (or occasionally shrunk) unless we further control the EoS during the estimation of the accuracy. We also provide several methods for controlling the EoS during the estimation of the accuracy, and a simple R package ("multisite.accuracy") that smoothly does this task for several accuracy estimates (e.g., sensitivity/specificity, area under the curve, correlation, hazard ratio, etc.).