节点文献

严重精神障碍者暴力行为预测模型应用研究

Application of violence prediction model for patients with SMD

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 万巍峙; 杨琴; 曹若辰; 秦小荣; 谌霞灿; 杨蕊; 王紫烨; 刘浩; 胡峻梅;

【Author】 Wan Weizhi;Yang Qin;Cao Ruochen;Qin Xiaorong;Shen Xiacan;Yang rui;Wang Ziye;Liu hao;Hu Junmei;West China School of Basic Medical Sciences & Forensic Medicine, Sichuan University;The Fourth People’s Hospital of Chengdu;Chengdu Public Security Bureau;

【通讯作者】 胡峻梅;

【机构】 四川大学华西基础医学与法医学院法医精神病学教研室; 成都市第四人民医院; 成都市公安局;

【摘要】 目的 基于朴素贝叶斯算法(Naive Bayes,NB)构建成都社区严重精神障碍(Severe Mental Disorder,SMD)患者暴力行为预测模型,并探索其应用价值。方法 从成都市精神卫生防治管理信息系统中获取2017至2019年SMD患者个案管理数据52 601例及相应随访信息、应急处置记录,使用MacArthur社区暴力工具、修订版外显攻击行为量表对患者随访期间的暴力行为进行量化评估。采用单因素Logistics回归分析筛选暴力行为的关联因素。将样本按照68%、17%、15%的比例划分为训练集、验证集、测试集,基于NB建立SMD患者暴力行为的预测模型。运用5折分层交叉验证法检验模型的内部效度,将测试集输入到模型中检验模型的外部效度。结果 基于NB建立的成都市社区SMD暴力行为预测模型内部效度的受试者工作特征曲线下面积(AUC)为0.757(95%CI:0.734~0.780),外部效度的AUC为0.755,平衡准确度为0.710,准确度为0.721,敏感度为0.697,特异度为0.722。结论基于NB建立的成都市社区SMD患者暴力行为预测模型具有良好的效度,可为暴力行为的预测提供新的方法和思路。

【Abstract】 Objective To establish a predictive model based on Naive Bayes(NB) for violent behavior of patients with severe mental disorder(SMD) in Chengdu community and explore its application value. Methods The case management data of 52,601 SMD patients from 2017 to 2019 from Chengdu Mental Health Prevention and Control Management Information System was obtained, including follow-up information on patients as well as emergency disposition record. MacArthur community violence instrument and Modified Overt Aggression Scales were used to evaluate the violent behavior of SMD patients during the follow-up period. Univariate logistics regression analysis was used to screen the associated factors of violent behavior. The data were divided into training set, validation set and test set at the proportions of 68 %, 17 % and 15 % respectively. The prediction model of violent behavior was established based on NB algorithm. The stratified 5-Fold cross-validation was used to test the internal validity of the model. The test set into the model to verify the external validity of the model was input. Results For the NB-based SMD violence prediction model, the area under the receiver operating characteristic curve(AUC) of the internal validity was 0.757(95 % CI: 0.734~0.780). AUC of external validity was 0.755, balance accuracy was 0.710, accuracy was 0.721, sensitivity was 0.697, specificity was 0.722, positive predictive value was 0.024, negative predictive value was 0.996. Conclusion The NB-based prediction model of SMD patients’ violence in Chengdu has good validity, which can provide new methods and ideas for the prediction of violent behavior.

【基金】 成都市科学技术局技术创新研发项目资助(2019-YF05-00240-SN);中国博士后科学基金第2批特别资助(站前)(2020TQ0219)
  • 【文献出处】 中国法医学杂志 ,Chinese Journal of Forensic Medicine , 编辑部邮箱 ,2022年04期
  • 【分类号】D919
  • 【下载频次】146
节点文献中: 

本文链接的文献网络图示:

本文的引文网络