• 中国核心期刊数据库收录期刊
  • 中文科技期刊数据库收录期刊
  • 中国期刊全文数据库收录期刊
  • 中国学术期刊综合评价数据库统计源期刊等

快速检索引用检索图表检索高级检索

中国医药导刊 ›› 2025, Vol. 27 ›› Issue (3): 271-276.

• 临床医药 • 上一篇    下一篇

基于增强回归树的无症状脑梗死影响因素分析

王佳芮12, 孔祥锴1, 赵素平1, 刘艳丽1, 杜娟12, 张晓蕴1*   

  1. 1.解放军总医院第九医学中心神经内科,北京 100101;
    2.安徽医科大学解放军306临床学院/安徽医科大学第五临床医学院,北京 100101
  • 收稿日期:2024-11-11 修回日期:2025-02-19 出版日期:2025-03-28 发布日期:2025-03-28
  • 基金资助:

    军队后勤自主项目(ZZCWS23J1005)

Analysis of Risk Factors for Silent Brain Infarction in Young and Middle-Aged Adults Based on Boosted Regression Tree Model

  1. 1.Department of Neurology the Ninth Medical Center of PLA General Hospital Beijing 100101, China
    2.306th Clinical College of PLA the Fifth Clinical College Anhui Medical University Beijing 100101, China
  • Received:2024-11-11 Revised:2025-02-19 Online:2025-03-28 Published:2025-03-28

摘要:

目的:探讨无症状脑梗死(silent brain infarctionSBI)的危险因素,并构建基于增强回归树的无症状脑梗死风险预测模型,为脑梗死的一级预防提供参考。方法:采用横断面研究,纳入201981日至202451日在我院健康管理科进行健康体检并接受头颅核磁检查的162例年龄40~60岁体检者,收集一般资料、调查问卷、血常规、生化及代谢相关的67项检测结果。采用增强回归树模型建立无症状脑梗死预测模型。结果:发生无症状脑梗死组61例,未发生无症状脑梗死组101例。增强回归树模型提示,低密度脂蛋白、体重指数、内中膜增厚、空腹血糖、凝血酶原活动度、尿素、总胆红素、肌酐是与无症状脑梗死发生风险最相关的因素,贡献度大于5%。模型的AUC值为0.983,提示模型预测准确性好。结论:构建的增强回归树模型对无症状脑梗死的发生具有良好的预测能力,可以识别出更符合临床意义的危险因素。

 

关键词:  , 无症状脑梗死;中青年;增强回归树;逻辑回归模型;相关风险因素

Abstract:

Objective: To investigate the impact factors associated with silent brain infarction SBI), and build a boosted regression tree model to predict SBI. The model may provide scientific reference for the primary prevention of brain infarction.Methods: In this cross-sectional study we enrolled 162 subjects in the Health Management Department of our medical center from August 1 2019 to May 1 2024. All the subjects aged between 40 and 60 and underwent physical examination and head MRI examination. General data questionnaire blood routine indexes 67 laboratory biochemical indexes related to biochemistry and metabolism were collected. The boosted regression tree model was used to establish the prediction model of asymptomatic cerebral infarction.Results: There were 61 cases in the SBI group and 101 cases in the non-SBI group. The boosted regression tree model suggested that low density lipoprotein body mass index intimal medial thickness fasting blood glucose prothrombin activity urea total bilirubin and creatinine were the most related factors to the risk of asymptomatic cerebral infarctionand their contribution was greater than 5%. The AUC of the model was 0.983 indicating that the model has good prediction accuracy.Conclusion: The boosted regression tree model has good predictive ability for the occurrence of SBI. The model can identify risk factors that are more in line with clinical significance and have certain clinical application value.


Key words:  , Silent brain infarction , Young and middle-aged people , Boosted regression tree , Logistic regression model , Related risk factors

中图分类号: