中国血液净化

中国血液净化 ›› 2024, Vol. 23 ›› Issue (07): 529-533.doi: 10.3969/j.issn.1671-4091.2024.07.009

• 临床研究 • 上一篇    下一篇

基于Lasso-Nomogram模型构建维持性血液透析患者睡眠障碍的预测模型

孙海云   尹沛然    钱 鹏   

  1. 215000 苏州，苏州大学附属第二医院1血液净化中心 2肾内科
  • 收稿日期:2023-11-23 修回日期:2024-04-10 出版日期:2024-07-12 发布日期:2024-07-12
  • 通讯作者: 钱鹏 E-mail:71815293@qq.com
  • 基金资助:
    苏州市科技计划项目(SYS2020132)

Construction of a prediction model for sleep disorders in maintenance hemodialysis patients based on Lasso-Nomogram model and verification of the model

SUN Hai-yun, YIN Pei-ran, QIAN Peng   

  1. Department of Nephrology, The Second Affiliated Hospital of Soochow University, Suzhou 215000, China
  • Received:2023-11-23 Revised:2024-04-10 Online:2024-07-12 Published:2024-07-12
  • Contact: 215000 苏州，苏州大学附属第二医院 1血液净化中心 E-mail:71815293@qq.com

PDF

4

摘要： 目的  基于Lasso-Nomogram模型构建维持性血液透析(maintenance hemodialysis，MHD)患者睡眠障碍(sleep disorder，SD)的预测模型。 方法  选取苏州大学附属第二医院行MHD的慢性肾衰竭(chronic renal failure，CRF)患者，根据MHD后6个月是否发生SD分为SD组和非SD组。比较2组临床资料，分析SD发生的影响因素，根据预测因素构建SD的Nomogram预测模型。 结果  198例CRF患者MHD后第6个月92例患者发生SD，SD发生率为46.46%；Logistic分析显示年龄(OR=2.152，95% CI：1.246～3.718，P＜0.001)、皮肤瘙痒(OR=6.209，95% CI：2.051～18.796，P＜0.001)、抑郁(OR=3.715，95% CI：1.531～9.013，P＜0.001)、尿素清除指数(urea clearance index，Kt/V)(OR=0.302，95% CI：0.154～0.592，P＜0.001)、血磷(OR=2.274，95% CI：1.236～4.185，P＜0.001)、钙磷乘积(OR=3.210，95% CI：1.517～6.792，P＜0.001)、血清合肽素(OR=6.816，95% CI：2.317～20.048，P＜0.001)、α-淀粉酶(OR=5.277，95% CI：1.953～14.257，P＜0.001)、25羟维生素D3(OR=0.381，95% CI：0.186～0.780，P＜0.001)均为SD发生的影响因素；根据Lasso、Logistic分析筛选出上述9个指标构建SD的Nomogram预测模型，该模型预测MHD患者发生SD的曲线下面积(AUC)为0.928(95% CI：0.892～0.963)，预测敏感度、特异度分别为81.13%、90.11%。 结论  根据MHD患者发生SD的因素构建Nomogram预测模型，在预测SD发生风险方面具有较高预测效能和良好临床效用。

关键词: 慢性肾衰竭, 维持性血液透析, 睡眠障碍, 影响因素, 预测模型

Abstract: Objective  To construct a prediction model of sleep disorder (SD) in patients with maintenance hemodialysis (MHD) based on Lasso-Nomogram model, and to verify the efficacy of the prediction model.  Methods   A total of 198 patients with chronic renal failure (CRF) who underwent MHD in our hospital were selected and categorized into SD and non-SD groups according to whether SD occurred 6 months after MHD. We compared the clinical data of the two groups, analyzed the influencing factors for SD, and constructed a nomogram prediction model of SD according to the predictive factors.  Results   In the sixth month after MHD, 92 CRF patients developed SD, with the SD incidence of 46.46% (92/198). Logistic analysis showed that age (OR=2.152, 95% CI：1.246～3.718), skin itching (OR=6.209, 95% CI：2.051～18.796), depression (OR=3.715, 95% CI：1.531～9.013), urea clearance index (Kt/V) (OR=0.302, 95% CI：0.154～0.592), blood phosphorus (OR=2.274, 95% CI：1.236～4.185), calcium and phosphorus product (OR=3.210, 95% CI：1.517～6.792), serum copeptin (OR=6.816, 95% CI：2.317～20.048), α-amylase (OR=5.277, 95% CI：1.953～14.257), and 25-(OH)D3 (OR=0.381, 95% CI：0.186～0.780) were the influencing factors for SD (P<0.001). A nomogram prediction model of SD was constructed based on the nine indicators screened by Lasso and logistic analyses. Using this model, the area under the curve (AUC) for the occurrence of SD in CRF patients with MHD was 0.928 (95% CI：0.892～0.963), with the prediction sensitivity and specificity of 81.13% and 90.11% respectively.  Conclusion  This nomogram prediction model of SD in CRF patients with MHD based on the influencing factors for SD has higher predictive efficacy and better clinical effect in predicting SD risk.

Key words: Chronic renal failure, Maintenance hemodialysis, Sleep disorders, Influencing factor, Prediction model

中图分类号: