中国血液净化

中国血液净化 ›› 2025, Vol. 24 ›› Issue (09): 786-790.doi: 10.3969/j.issn.1671-4091.2025.09.013

• 护理研究 • 上一篇    下一篇

维持性血液透析患者症状群困扰的影响因素分析及决策树模型的建立与评价

王 玲   朱亚梅   吕小林   张 晶   邵 郁   顾 月   

  1. 210029 南京，1南京医科大学第一附属医院(江苏省人民医院)肾内科
  • 收稿日期:2024-09-30 修回日期:2025-06-19 出版日期:2025-09-12 发布日期:2025-09-12
  • 通讯作者: 朱亚梅 E-mail:zymei6868@126.com

Analysis of the influencing factors for symptom group disturbance in maintenance hemodialysis patients and establishment and evaluation of the decision tree model

WANG Ling, ZHU Ya-mei, LYU Xiao-lin, ZHANG Jing, SHAO Yu, GU Yue   

  1. Department of Nephrology, The First Affiliated Hospital of Nanjing Medical University (Jiangsu Provincial People's Hospital), Nanjing 210029, China
  • Received:2024-09-30 Revised:2025-06-19 Online:2025-09-12 Published:2025-09-12
  • Contact: 210029 南京，1南京医科大学第一附属医院(江苏省人民医院)肾内科 E-mail:zymei6868@126.com

摘要： 目的  对维持性血液透析(maintenance hemodialysis，MHD)患者症状群困扰的影响因素进行分析，并构建决策树风险预测模型。 方法  回顾性选取2019年3月—2024年3月南京医科大学第一附属医院收治的MHD患者作为研究对象，收集所有患者临床相关资料，根据是否发生症状群困扰分为困扰组和非困扰组。使用单因素及多因素Logistic回归分析影响MHD患者发生症状群困扰的危险因素，采用决策树模型与Logistic回归算法构建其风险预测模型，并比较2种模型对MHD患者发生症状群困扰的预测价值。结果  共纳入398例患者，其中困扰组106例、非困扰组292例。多因素Logistic分析显示：家庭月收入(OR=2.072，95% CI：1.019～4.211，P=0.044)、透析龄(OR=2.046，95% CI：1.055～3.970，    P=0.034)、焦虑抑郁(OR=1.990，95% CI：1.002～3.949，P=0.049)、甲状旁腺激素(OR=1.241，95% CI：1.178～1.308，P＜0.001)和血磷(OR=18.581，95% CI：5.963～57.897，P＜0.001)是MHD患者并发症状群困扰的危险因素，血红蛋白(OR=0.947，95% CI：0.925～0.969，P＜0.001)是MHD患者并发症状群困扰的保护因素。构建的决策树模型包含5个变量，分为4层共24个节点，甲状旁腺激素是主要影响因素。   Logistic回归模型和决策树模型的AUC分别为0.898和0.923，两者比较的delong检验Z=2.240，     P=0.025。 结论  家庭月收入、透析龄、焦虑抑郁、甲状旁腺激素和血磷是MHD患者症状群困扰的危险因素，血红蛋白是保护因素。构建的决策树风险预测模型预测效能优于Logistic回归模型。

关键词: 维持性血液透析, 症状群, 影响因素, 决策树, 风险预测模型

Abstract: Objective To analyze the influencing factors for symptom cluster distress in maintenance hemodialysis (MHD) patients and to construct a decision tree risk prediction model.  Methods A total of 398 MHD patients admitted to the First Affiliated Hospital of Nanjing Medical University from March 2019 to March 2024 were retrospectively reviewed as the study subjects. Their clinical data were recruited. They were then categorized into distress and non-distress groups according to the presence or absence of symptom cluster distress. The risk factors affecting the symptom cluster distress in MHD patients were analyzed using unifactorial and multifactorial logistic regressions. The risk prediction models were constructed using the decision tree model and logistic regression algorithm, and the predictive value of the 2 models for the occurrence of symptom cluster distress in MHD was compared.  Results A total of 398 patients were included, including 106 cases in distress group and 292 cases in non-distress group. Multifactorial logistic analysis showed that monthly income (OR=2.072, 95% CI:1.019～4.211, P=0.044), dialysis time (OR=2.046, 95% CI:1.055～3.970, P=0.034), anxiety and depression (OR=1.990, 95% CI:1.002～3.949, P=0.049), parathyroid hormone (OR=1.241, 95% CI:1.178～1.308, P＜0.001) and blood phosphorus (OR=18.581, 95% CI:5.963～57.897, P＜0.001) were the risk factors for the presence of symptom cluster distress in MHD patients, and hemoglobin (OR=0.947, 95% CI:0.925～0.969, P＜0.001) was a protective factor for the presence of symptom cluster distress in MHD patients. The constructed decision tree model contained 5 variables that divided into 4 levels with 24 nodes, and parathyroid hormone was the main influencing factor. The AUCs of logistic regression model and decision tree model were 0.898 and 0.923, respectively, and the delong test for comparison of the two models was Z=2.240 and P=0.025.  Conclusion   Monthly income, dialysis time, anxiety and depression, parathyroid hormone and blood phosphorus are the risk factors for symptom cluster distress in MHD patients, and hemoglobin is a protective factor. The predictive efficacy of the constructed decision tree risk prediction model was superior to that of the logistic regression model. 

Key words: Maintenance hemodialysis, Symptom cluster, Influencing factor, Decision tree, Risk prediction model

中图分类号: