成人脑卒中患者心理痛苦风险预测模型的构建与验证  

Construction and validation of a prediction model of psychological distress risk in adult patients with stroke

作　　者：陈雁南 庄淑梅[1] 靳世梅 周雪莹 CHEN Yannan;ZHUANG Shumei;JIN Shimei;ZHOU Xueying(Nursing College,Tianjin Medical University,Tianjin 300070,China)

机构地区：[1]天津医科大学护理学院,天津300070

出　　处：《天津医科大学学报》2025年第2期144-149,共6页Journal of Tianjin Medical University

基　　金：天津市深化医药卫生体制改革研究项目立项课题(2022-YG06)。

摘　　要：目的:构建成人脑卒中患者心理痛苦风险预测模型,并验证其预测效果。方法:选取2023年2月至10月天津医科大学总医院、天津市第一中心医院、天津市宝坻区人民医院及武警特色医学中心神经科的脑卒中患者428例,采用一般资料调查问卷、心理痛苦温度计(DT)、疼痛数字评价量表(NRS)、一般自我效能感量表(GSES)、领悟社会支持量表(PSSS)进行调查,采用单因素分析和二元Logistic回归分析筛选心理痛苦的危险因素,建立Logistic回归预测模型和随机森林预测模型。结果:Logistic回归分析表明年龄、脑卒中类型、NRS得分、GSES得分和PSSS得分是心理痛苦的独立危险因素(OR=1.146、22.435、4.953、0.713、0.920,均P<0.05)。Logistic回归模型与随机森林模型的预测准确率、痛苦准确率和受试者工作特征(ROC)曲线下面积(AUC)值分别为93.0%和94.5%、66.7%和87.5%、0.961(0.940~0.982)和0.886(0.789~0.982)。结论:Logistic回归模型与随机森林模型均能预测成人脑卒中患者的心理痛苦,而随机森林模型对于存在明显心理痛苦患者的识别能力和预测效果均优于Logistic回归模型。Objective:To construct a prediction model of psychological distress risk in adult patients with stroke and validate its prediction effect.Methods:A total of 428 adult stroke patients from the General Hospital of Tianjin Medical University,the First Central Hospital of Tianjin,the People′s Hospital of Tianjin Baodi District and the Special Medical Center of the Armed Police Force were selected from February to October 2023.A self-designed general data questionnaire,Distress Thermometer(DT),Numerical pain Rating Scale(NRS),General Self-Efficacy Scale(GSES),and Perceptive Social support Scale(PSSS)were used for the survey.Univariate analysis and binary Logistic regression analysis were employed to identify risk factors associated with psychological distress.Logistic regression prediction model and random forest prediction model were established.Results:Logistic regression analysis showed that age,stroke type,NRS score,GSES score and PSSS score were independent risk factors of mental distress(OR=1.146,22.435,4.953,0.713,0.920,all P<0.05).The prediction accuracy,psychological distress accuracy and area under the receiver operating characteristic(ROC)curve(AUC value)of the Logistic regression model and the random forest model were 93.0%and 94.5%,66.7%and 87.5%,0.961(0.940-0.982)and 0.886(0.789-0.982),respectively.Conclusion:Both the Logistic regression model and the random forest model can predict psychological distress in adult stroke patients,while the random forest model is better than the Logistic regression model in identifying and predicting patients with significant psychological distress.

关 键 词：脑卒中 心理痛苦 预测模型 随机森林 

分 类 号：R743.3[医药卫生—神经病学与精神病学]