The Role of Social Determinants in Forensic Outcomes of Schizophrenia Spectrum Disorders: A Machine Learning-Supported Cohort Study

Abstract

Maintaining a healthy social network is a pervasive challenge among individuals with schizophrenia spectrum disorders (SSD), profoundly influencing mental health outcomes and social functioning. While social adversities such as social isolation may not be a direct cause of criminal behavior, they may act as a significant risk factor. Previous results from offenders with SSD indicated that attention disorder, alogia, crime motivated by ego disturbances, overall symptom severity, and a history of negative symptoms were the most influential factors in predicting social isolation in this population. The present study aims to examine whether specific social determinants can distinguish between forensic and non-forensic patients with SSD, thereby addressing the research question of whether these factors are associated with an increased risk of offending in this population. The comparative pattern analysis was conducted using a variety of machine learning models. A history of social isolation, relationship status upon discharge from the institution, social marginalization during childhood, complete compulsory schooling, and contact with the mental health services through family members demonstrated the highest capacity to distinguish between these two groups. Factors related to psychopathology however were statistically outperformed in the prediction model, which yielded an AUC of 0.83. The study’s results highlight determinants of offending related to social integration/isolation as an outcome of SSD and possibly indicating potential preventative measures regarding offending SSD patients.