Prediction of Violent Behavior in Patients With Schizophrenia by Multimodal Machine Learning

Overview

Patients with schizophrenia have a higher risk of committing violent crimes than the general population, and the relative risk of violence against others is four times higher than the general population. Violence is a major public health problem because it often leads to poor prognosis, readmission and stigma in patients with schizophrenia. MRI studies on violent behavior in schizophrenia are relatively few. These studies have found that violence is primarily associated with dysfunction in the ventral prefrontal and temporal limbic systems. Structural MRI found that violent behavior in schizophrenia was associated with increased volume of white matter in caudate nucleus, left orbitofrontal gyrus and right orbitofrontal gyrus. However, the current research results in this field are uneven, the methods are not consistent, and there is a lack of breakthrough progress, which needs to be integrated and deepened urgently. If the violent behavior of the patients with schizophrenia could be predicted by magnetic resonance imaging, it would be a revolutionary try. By doing so, the investigators can strengthen the treatment of these patients and reduce the occurrence of violence. Based on previous studies, the investigators believe that violent schizophrenics exhibit recognizable imaging characteristics under structural phase, resting state, negative emotional images and natural stimuli models. Anomalies in a particular mode may be subtle and difficult to identify, but when multiple different modes are integrated, a significant and characteristic set of imaging markers will be present. This study will use the multivariate model of machine learning method, detection brain activation patterns under different situations among patients with violence. The investigators are going to study imaging biomarkers, and try to predict the possibility of onset of violence among schizophrenia patients, thus reduce the risks of violence.

Conditions