基于列线图构建社区男性精神分裂症患者发生暴力犯罪行为的预测模型

Construct a prediction model of violent criminal behavior of male schizophrenics in community based on the line graph

目的研究基于列线图构建社区男性精神分裂症患者发生暴力犯罪行为的预测模型的价值。方法分析2021年9月—2022年3月于宜春市第三人民医院接受诊治的97例男性精神分裂症患者临床资料。对所有受试者均进行为期6个月的随访观察,根据是否发生暴力犯罪行为,将受试者分作暴力犯罪行为组(30例)以及未发生暴力犯罪行为组67例。比较2组各项基线资料,并采用单因素及多因素分析精神分裂症患者发生暴力犯罪行为的影响因素。经R软件建立预测暴力犯罪行为的列线图模型,绘制ROC曲线验证该模型区分度。结果暴力犯罪行为组饮酒史、未遵医嘱用药、既往暴力史人数占比均高于未发生暴力犯罪行为组;且阳性与阴性症状量表(PANSS)评分高于未发生暴力犯罪行为组,而SSRS评分低于未发生暴力犯罪行为组(P<0.05)。经多因素Logistic回归分析发现,饮酒史、未遵医嘱用药、既往暴力史及PANSS评分均是社区男性精神分裂症患者发生暴力犯罪行为的危险因素(P<0.05)。经ROC曲线分析发现,列线图模型预测社区男性精神分裂症患者发生暴力犯罪行为风险的曲线下面积为0.833(95%CI为0.759~0.892)。结论基于列线图构建社区男性精神分裂症患者发生暴力犯罪行为的预测模型具有一定价值。

Objective The value of constructing of a predictive model of violent crime in community male schizophrenia patients based on nomogram.Methods The clinical data of 97 male patients with schizophrenia who were diagnosed and treated in Yichun Third People's Hospital from September 2021 to March 2022 were analyzed.All subjects were followed up for 6 months,and they were divided into 30 cases of violent crime group and 67 cases of no violent crime group according to whether they had violent crime.The baseline data of the two groups were compared,and the influencing factors of violent crime in patients with schizophrenia were analyzed by univariate and multivariate analysis.The nomogram model for predicting violent crime was established by R software,and the ROC curve was drawn to verify the discrimination of the model.Results The proportions of the number of people with a history of drinking,non-compliance with medication,and history of violence in the violent crime group were higher than those in the non-violent crime group,the PANSS score was higher than that in the non-violent crime group,and the SSRS score was lower than that in the non-violent crime group Criminal behavior group(P<0.05).Multivariate Logistic regression analysis showed that drinking history,non-compliance with medication,previous violent history and PANSS score were all risk factors for violent crime in community male schizophrenia patients(P<0.05).The ROC curve analysis showed that the area under the curve of the nomogram model for predicting the risk of violent criminal behavior among male schizophrenia patients in the community was 0.833(95%CI 0.759-0.892).Conclusion It is of certain value to construct a predictive model of violent crime in community male schizophrenia patients based on nomogram.