Early Identification of Patients Presenting a First Psychotic Episode, Non Answering to First Line Support Strategies

Recently, radiological markers demonstrated a good capacity to discriminate first episode of psychosis patients, that will show a bad response to treatment. Furthermore, automatic learning methods were recently successfully applied to neurological data, but never for the first episode psychosis patients. These powerful methods are based on multivariate analyses allowing the prediction at an individual stage. The possibility to identify more precisely the population of non-answering patients would allow the adaptation of the support in first attention for these patients, and consequently to perform appropriate cares. Researchers propose in this project to evaluate predictive factors of non response. It is defined by the lack of symptomatic remission at month 3, for patients presenting a first episode of psychosis. It will be performed on objective markers taken from imagery techniques. The link between evolution of these markers and clinical measures will be studied.