Hudon Alexandre, Dellazizzo Laura, Phraxayavong Kingsada, Potvin Stéphane, Dumais Alexandre
Services et Recherches Psychiatriques AD.
J Nerv Ment Dis. 2023 Feb 1;211(2):88-94. doi: 10.1097/NMD.0000000000001604.
The objective of this cross-sectional study was to identify cannabis-related features and other characteristics predictive of violence using a data-driven approach in patients with severe mental disorders (SMDs). A Least Absolute Shrinkage and Selection Operator regularization regression model was used on the database consisting of 97 patients with SMD who completed questionnaires measuring substance use and violence. Cannabis use, particularly related to patients' decision to consume or time spent using, was a key predictor associated with violence. Other patterns of substance use and personality traits were identified as strong predictors. Regular patterns of cannabis use and interpersonal issues related to cannabis/stimulant abuse were inversely correlated to violence. This study identified the effect of several predictors correlated to violence in patients with SMD using a regularization regression model. Findings open the door to better identify the profiles of patients that may be more susceptible to perpetrate violent behaviors.
这项横断面研究的目的是采用数据驱动的方法，在重度精神障碍（SMD）患者中识别与大麻相关的特征以及其他预测暴力行为的特征。对由97名完成了测量物质使用和暴力行为问卷的SMD患者组成的数据库，使用了最小绝对收缩和选择算子正则化回归模型。大麻使用，特别是与患者的消费决定或使用时间有关的使用，是与暴力行为相关的关键预测因素。其他物质使用模式和人格特质也被确定为强有力的预测因素。大麻的规律使用以及与大麻/兴奋剂滥用相关的人际问题与暴力行为呈负相关。本研究使用正则化回归模型确定了与SMD患者暴力行为相关的几个预测因素的影响。研究结果为更好地识别可能更容易实施暴力行为的患者特征打开了大门。
