实用医学杂志 ›› 2026, Vol. 42 ›› Issue (8): 1471-1478.doi: 10.3969/j.issn.1006-5725.2026.08.023

• 论著·临床实践 • 上一篇    

比较决策树与logistic回归模型对老年股骨粗隆间骨折患者术后脑梗死风险的预测效能

吴红玉1,2,吴昱芃1,黄少娟1,3(),黄素珍1,2,侯英杰1,2,杨毅1,2,张志荣1,2,李庆锡1,4,吴智鑫1,5   

  1. 1.广州中医药大学第八临床医学院(广东佛山 528000)
    2.佛山市中医院 手术中心
    3.佛山市中医院护理部
    4.佛山市中医院病案室
    5.佛山市中医院 神经重症监护病房 (广东 佛山 528000 )
  • 收稿日期:2025-12-05 出版日期:2026-04-25 发布日期:2026-04-28
  • 通讯作者: 黄少娟 E-mail:flamewu@163.com
  • 基金资助:
    广东省中医药局科研项目(20241307);广东省基础与应用基础研究基金项目(粤佛地区培育项目)(2022A1515140060)

Risk prediction of postoperative cerebral infarction in elderly patients with intertrochanteric femoral fracture: A comparative study of decision tree and logistic regression model

Hongyu WU1,2,Yupeng WU1,Shaojuan HUANG1,3(),Suzhen HUANG1,2,Yingjie HOU1,2,Yi YANG1,2,Zhirong ZHANG1,2,Qingxi LI1,4,Zhixin WU1,5   

  1. 1.The Eighth Clinical Medical College,Guangzhou University of Chinese Medicine,Foshan 528000,Guangdong,China
    2.Surgery Center,Foshan Hospital of Traditional Chinese Medicine,Foshan 528000,Guangdong,China
    3.Nursing Department,Foshan Hospital of Traditional Chinese Medicine,Foshan 528000,Guangdong,China
    4.Medical Records Room,Foshan Hospital of Traditional Chinese Medicine,Foshan 528000,Guangdong,China
    5.Neuro?Intensive Care Unit,Foshan Hospital of Traditional Chinese Medicine,Foshan 528000,Guangdong,China
  • Received:2025-12-05 Online:2026-04-25 Published:2026-04-28
  • Contact: Shaojuan HUANG E-mail:flamewu@163.com

摘要:

目的 对老年股骨粗隆间骨折(intertrochanteric fracture of the femur,IFF)术后合并脑梗死的危险因素进行分析,并建立决策树风险预测模型,为临床医务人员制定针对性的防治方案提供科学的理论依据。 方法 回顾性选取2017年11月至2023年12月佛山市中医院收治的215例行股骨粗隆骨折手术治疗的老年患者临床相关资料,根据术后是否发生脑梗死,将其分为脑梗死组和非脑梗死组,采用单因素及多因素logistic回归分析筛选出影响老年IFF患者术后并发脑梗死的危险因素,采用决策树模型与logistic回归构建其风险预测模型,并比较两种模型对老年IFF患者术后并发脑梗死的预测价值。 结果 215例手术治疗的老年IFF患者,术后共有61例发生脑梗死,其发生率为28.37%;多因素logistic回归分析结果显示,高血压、糖尿病、住院卧床时间、同型半胱氨酸(Hcy)以及D-二聚体(D-D)为老年IFF患者术后合并脑梗死的危险因素,优质护理为其保护因素(P < 0.05);根据危险因素生成决策树模型,模型选择了糖尿病、住院卧床时间、Hcy、D-D以及优质护理等5个解释变量,合计4层,共15个节点,其中住院卧床时间为老年IFF患者术后合并脑梗死最为重要的影响因素;老年IFF患者术后合并脑梗死的决策树模型曲线下面积(AUC)值为0.964(95%CI:0.930 ~ 0.985),logistic回归模型AUC值为0.896(95%CI:0.847 ~ 0.933),两种模型的delong检验结果为Z = 3.401,P = 0.000 7。 结论 高血压、糖尿病、住院卧床时间、Hcy以及D-D为老年IFF患者术后合并脑梗死的危险因素,优质护理为其保护因素,根据危险因素构建的决策树风险预测模型,其预测效能显著高于logistic回归模型。

关键词: 决策树模型, logistic回归模型, 老年股骨粗隆间骨折, 脑梗死, 影响因素

Abstract:

Objective To analyze the risk factors for cerebral infarction after surgery for intertrochanteric fracture of the femur (IFF) among elderly patients and establish a decision-tree risk prediction model, so as to provide a scientific theoretical basis for clinical medical staff to formulate targeted prevention and treatment plans. Methods Clinical data of 215 elderly patients who underwent surgery for IFF at our hospital from November 2017 to December 2023 were retrospectively selected. The patients were divided into the cerebral infarction group and the non-cerebral infarction group according to the postoperative occurrence of cerebral infarction. Univariate and multivariate logistic regression analyses were conducted to identify the risk factors for cerebral infarction after IFF surgery in elderly patients. A decision tree model and logistic regression were utilized to construct a risk prediction model, and the predictive values of both models for cerebral infarction after IFF surgery were compared. Results Among the 215 elderly patients who underwent surgery for IFF, 61 suffered from cerebral infarction post-operatively, with an incidence rate of 28.37%. Multivariate logistic regression analysis indicated that hypertension, diabetes, the duration of hospital bed rest, homocysteine, and D-dimer were risk factors for cerebral infarction after IFF surgery in elderly patients, while quality nursing was identified as a protective factor (P < 0.05). A decision tree model was constructed based on these risk factors, with diabetes, the duration of hospital bed rest, Hcy, D-D, and quality nursing selected as explanatory variables. The model had a total of 4 layers and 15 nodes. The duration of hospital bed rest was the most significant influencing factor for cerebral infarction after IFF surgery in elderly patients. The area under the curve (AUC) value of the decision tree model for cerebral infarction after IFF surgery in elderly patients was 0.964 (95%CI: 0.930 - 0.985), whereas the AUC value of the logistic regression model was 0.896 (95%CI: 0.847 - 0.933). The DeLong test result for the two models was Z = 3.401, P = 0.000 7. Conclusions Hypertension, diabetes, the duration of hospital bed rest, Hcy, and D-D are identified as risk factors for cerebral infarction after IFF surgery in elderly patients, while quality nursing is recognized as a protective factor. The decision tree risk prediction model constructed based on these risk factors demonstrates significantly higher predictive efficacy compared to the logistic regression model.

Key words: decision tree model, logistic regression model, elderly intertrochanteric femoral fracture, cerebral infarction, influencing factors

中图分类号: