UP-TO-DATE RESEARCH – A Machine Learning approach for Violent Behaviors and Stress in Schizophrenia

Category: News

Situation to date: 

One of the main aims of the EU-VIORMED Project is to identify risk factors for violence amongst individuals with mental disorder. Previous studies have shown that stress is an important link in the chain of triggers of violence. Nonetheless, few studies have investigated whether stress can explain, at least partly, the relationship between mental disorders and offending. No study to date has inquired this relationship in people suffering from schizophrenia spectrum disorders (SSDs). Indeed, violent behavior in patients with SSD is often attributed to the disorder itself, without a closer consideration of the situational and interpersonal factors. The following study deals with the potential role of stressful experiences.

The study:

A recent study (April 2020) published in the Journal of Interpersonal Violence (Impact Factor 3.064) by Johannes Kirchebner and colleagues, aimed at assessing how accumulation and type of stressors influence the severity of offenses perpetrated by patients suffering from an SSD. The investigators examined a sample of 370 offenders who were hospitalized in the Centre for Inpatient Forensic Therapy at the Zurich University Hospital of Psychiatry from 1982 to 2016 with a diagnosis of an SSD, reviewing the subjects’ available records. Using logistic regression and machine learning (ML) algorithms, the researchers explored whether a number of past stressors were associated with a higher likelihood of committing a violent offense in order to identify those potentially explaining the violent behavior. Main findings suggested that an accumulation of stressful experiences in the lifetime correlates with the likelihood of committing violent offenses, consistently with the results of previous studies in people without psychotic disorders. Assessing the types of stressors, the researchers identified some key stressors: for instance, previous coercive psychiatric treatments, unemployment, and separation from main caregivers in childhood and/or youth appeared as predictive factors of violent offenses in individuals with SSDs. On the other hand, social isolation in adulthood and failure in school did not seem to lead to a higher probability of committing a violent offense.

Take-home message:

Despite some limitations, these findings support the use of ML algorithms to expand the current knowledge about the relationship between stress and violent offending in patients with SSDs, which is crucial for the development of preventive and therapeutic strategies. These may potentially reduce the prevalence of violent offenses in this population.

Further research is needed to gain more insights into SSDs and offending.


The article full text is available at:

https://journals.sagepub.com/doi/10.1177/0886260520913641


Full reference:

Kirchebner J, Sonnweber M, Nater UM, Günther M, Lau S. Stress, Schizophrenia, and Violence: A Machine Learning Approach. J Interpers Violence; published online: April 20, 2020. doi: 10.1177/0886260520913641

Read more news...