摘要
目的探讨并比较基于Logistic回归模型和XGBoost模型预测老年脑卒中患者发生护理依赖的风险因素及预测性能。方法采用一般资料调查表、护理依赖量表、视觉模拟评分法、Barthel指数、微型营养评估量表和一般自我效能感量表对某医院收治的403例老年脑卒中患者进行调查,根据护理依赖发生情况分为发生组(n=258)和未发生组(n=145)。采用Logistic回归分析老年脑卒中患者发生护理依赖的影响因素,应用Logistic回归模型和XGBoost模型评价模型的区分度、校准度和临床实用性。结果老年脑卒中患者护理依赖发生率为64.02%,多因素分析结果显示,性别、文化程度、职业状况、居住状况、慢性病数量、近期住院天数、是否使用辅助设备、近期跌倒史、疼痛程度、肢体功能依赖程度和自我效能感是老年脑卒中患者发生护理依赖的独立影响因素。XGBoost模型的曲线下面积、准确率、精确率、灵敏度、特异度和F1值均优于Logistic回归模型,校准性良好(P=0.243),且在0~0.4风险阈值范围内具有临床价值。特征重要性分析显示,影响老年脑卒中患者护理依赖的主要因素依次为肢体功能依赖程度、自我效能感、疼痛程度、居住状况和慢性病数量。结论老年脑卒中患者护理依赖状况有待改善,XGBoost模型能有效预测老年脑卒中患者护理依赖风险,应重点关注患者的肢体功能康复、心理支持、疼痛管理、社会支持和慢性病管理,以降低护理依赖的发生风险。
Objective To explore and compare the risk factors and prediction performance of nursing dependence in elderly stroke patients based on Logistic regression model and XGBoost model.Methods A general information questionnaire,Care Dependence Scale(CDS),Visual Analogue Scale(VAS),Barthel Index(BI),Mini Nutritional Assessment(MNA),and General Self-Efficacy Scale(GSES)were used to investigate a total of 403 elderly stroke patients admitted to a hospital.According to the occurrence of nursing dependence,they were divided into an occurrence group(n=258)and a non-occurrence group(n=145).Logistic regression was used to analyze the influencing factors of nursing dependence in elderly stroke patients.Logistic regression model and XGBoost model were used to evaluate the discrimination,calibration and clinical practicality of the model.Results The incidence of care dependence in elderly stroke patients was 64.02%.Multivariate analysis identified gender,education level,occupational status,living conditions,number of chronic diseases,recent hospitalization days,use of assistive devices,recent history of falls,pain level,limb function dependence,and self-efficacy as independent risk factors.The area under the curve,accuracy,precision,sensitivity,specificity and F 1 value of the XGBoost model were better than those of the Logistic regression model,with good calibration(P=0.243),and it had clinical value within the risk threshold range of 0~0.4.Feature importance analysis showed that the main factors affecting nursing dependence in elderly stroke patients were limb function dependence,self-efficacy,pain level,living conditions and number of chronic diseases.Conclusion The nursing dependence of elderly stroke patients needs to be improved.The XGBoost model can effectively predict the risk of care dependence in elderly stroke patients.Clinically,the focus should be on the rehabilitation of limb function,psychological support,pain management,social support,and chronic disease management to reduce the risk of care dependence.
作者
孙家蓉
严金秀
董优清
吴海滨
刘啟
SUN Jiarong;YAN Jinxiu;DONG Youqing;WU Haibin;LIU Qi(Department of Neurosurgery,The First Affiliated Hospital of Nanchang University,Nanchang 330006,China;Department of Anesthesiology,The First Affiliated Hospital of Nanchang University,Nanchang 330006,China)
出处
《护理管理杂志》
2025年第5期404-410,共7页
Journal of Nursing Administration
基金
江西省卫生健康委员会科技计划项目(202510261)。
