首页|慢性心力衰竭合并2型糖尿病患者易损期心源性死亡的中西医预警指标分析

慢性心力衰竭合并2型糖尿病患者易损期心源性死亡的中西医预警指标分析

Analysis of Chinese and Western warning indicators of cardiogenic death in chronic heart failure combined with type 2 diabetes mellitus during the vulnerable phase

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
目的:分析慢性心力衰竭合并2型糖尿病患者发生易损期心源性死亡的中西医预警指标,为临床实践提供循证医学证据.方法:将山东省7家中医院符合研究标准的727例患者纳入队列研究.运用Cox回归结合多重支持向量机-递归特征消除(MSVM-RFE)模型分析预警指标.结果:将Cox单因素回归分析获得的53个潜在预警指标按重要性排序依次纳入MSVM-RFE模型进行效能检验,当纳入前11位潜在预警指标时,获得最小错误率0.177与最大精度0.823,即为最优预警指标子集.Cox多因素回归分析显示,纽约心脏病学会心功能分级(Ⅲ级vsⅡ级:HR=2.33,Ⅳ级vsⅡ级:HR=2.59),吸烟史(HR=1.75),血浆氨基末端脑利钠肽前体(NT-ProBNP)(HR=8.08),气阴两虚血瘀证(HR=2.26),面色口唇紫暗(HR=3.63),脉细数无力或结代(HR=1.02),少苔或有裂纹(HR=1.05),空腹血糖(FBG)(HR=1.02)为损害性预警指标;益气类注射液(HR=0.44)、降糖类药物(HR=0.29),降血脂类药物(HR=0.96)为保护性预警指标.结论:Cox回归与MSVM-RFE模型相结合的数据分析方法可实现优势互补,适用于高维、多重共线性、小样本数据的中西医预警指标分析,具有临床推广价值.
Objective:To analyze the warning indicators of Chinese and Western medicine in patients with chronic heart failure combined with type 2 diabetes mellitus who develop cardiogenic death during the vulnerable phase,and to provide evidence-based medical evidence for clinical practice.Methods:A total of 727 patients from 7 hospitals in Shandong Province who met the study criteria were included in the cohort study.Cox regression combined with multiple support vector machine-recursive feature elimination(MSVM-RFE)model was used to analyze the warning indicators.Results:The 53 potential warning indicators obtained from the Cox univariate regression analysis were sequentially included in the MSVM-RFE model in order of importance for the effectiveness test,and when the top 11 potential warning indicators were included,the model obtained a minimum error rate of 0.177 and a maximum precision of 0.823,so they were used as the optimal subset of warning indicators.The Cox multifactorial regression analysis showed that the New York Heart Association(NYHA)cardiac function classification(Class Ⅲ vs Class Ⅱ:HR=2.33,Class Ⅳ vs Class Ⅱ:HR=2.59),smoking history(HR=1.75),plasma amino-terminal brain natriuretic peptide precursor(NT-ProBNP)(HR=8.08),qi and yin deficiency and blood stasis syndrome(HR=2.26),purple face and lips(HR=3.63),thin and weak pulse or knotted pulse(HR=1.02),fissured tongue with little coating(HR=1.05),fasting blood glucose(FBG)(HR=1.02)are damaging warning indicators.Yiqi injections,hypoglycemic drugs,and hypolipidemic drugs Yiqi injections(HR=0.44),hypoglycemic drugs(HR=0.29),hypolipidemic drugs(HR=0.96)are protective warning indicators.Conclusion:The combination of MSVM-RFE and Cox regression data analysis method could achieve complementary advantages.It is suitable for the analysis of Chinese and Western medicine warning indicators with high-dimensional,multi-collinearity and small sample data,and has clinical promotion value.

Chronic heart failureType 2 diabetes mellitusVulnerable phase of heart failureChinese and Western medicine warning indicatorsMultiple support vector machine-recursive feature eliminationCox regression

李润民、戴国华、高武霖、管慧、任丽丽、王兴蒙、曲惠文

展开 >

山东中医药大学,济南 250000

山东中医药大学附属医院,济南 250000

慢性心力衰竭 2型糖尿病 心衰易损期 中西医预警指标 多重支持向量机-递归特征消除 Cox回归

国家重点研发计划国家自然科学基金面上项目国家自然科学基金青年科学基金山东省自然科学基金青年基金中国博士后科学基金面上项目

2019YFC17104018217417282204942ZR2022QH1232022M721998

2024

中华中医药杂志
中华中医药学会

中华中医药杂志

CSTPCD北大核心
影响因子：1.135
ISSN：1673-1727
年,卷(期)：2024.39(6)
  • 12