基于中医体质量表和常规体检指标构建新发心脑血管事件的列线图预测模型*
作者：潘雨薇1,2，赵海娴2，高三德1，赖 婵3，杜文坚1，孙 晶4，蔡泽建5
单位：1.广州医科大学附属中医医院，广东 广州 510640； 2.广州市天河区中医医院，广东 广州 510630； 3.广州市前进街道社区卫生服务中心，广东 广州 510660； 4.广州市林和街道社区卫生服务中心，广东 广州 510630； 5.广州市猎德街道社区卫生服务中心，广东 广州 510663
引用：引用：潘雨薇，赵海娴，高三德，赖婵，杜文坚，孙晶，蔡泽建.基于中医体质量表和常规体检指标构建新发心脑血管事件的列线图预测模型[J].中医药导报,2025,31(7):79-85.
DOI：10.13862/j.cn43-1446/r.2025.07.013
PDF：
下载PDF
摘要：
目的：基于广州市基本公共卫生项目中医体质量表和常规体检指标，分析新发心脑血管事件（MACCE）的相关危险因素，建立相应列线图模型并评价其有效性。方法：纳入广东省广州市531名65~90岁老年人（MACCE组382人，非MACCE组149人），按7∶3比例将样本随机分为训练集（n=372）与验证集（n=159）。先经单因素Logistic回归分析筛选变量，再用LASSO回归确定建模因子，最后以多因素Logistic回归分析筛选MACCE独立危险因素并构建列线图模型。通过校准曲线、内部bootstrap验证、ROC曲线分析及验证集验证评估模型性能。结果：训练集单因素分析显示平和质、气虚质、糖尿病病史、高血压病史、甘油三酯升高和低密度脂蛋白升高差异有统计学意义（P＜0.05）。LASSO回归确定15个预测因子，多因素分析确定7个独立影响因素（气虚质、痰湿质、糖尿病病史、高血压病史、甘油三酯升高和低密度脂蛋白升高）构建列线图模型。校准曲线显示模型校准性能良好，训练集C指数为0.846，ROC值为0.835；验证集C指数为0.857，AUC值为0.844。结论：气虚质、痰湿质、糖尿病病史、高血压病史、甘油三酯升高和低密度脂蛋白升高是新发心脑血管事件的独立危险因素。本研究建立的列线图模型有良好的预测效能和区分度，可预测新发心脑血管事件的风险。
关键词：新发心脑血管事件；中医体质；体检指标；列线图预测模型
Abstract：
Objective:
Based on the Chinese Medicine Physical Examination Scale of the Guangzhou Basic
Public Health Program and the indicators of routine physical examination, to
analyze the risk factors associated with new-onset cardiovascular and
cerebrovascular events (MACCE), then to establish the corresponding columnar
graphical model, and evaluate its validity. Methods: A total of 531 elderly
people aged 65-90 years (382 in the MACCE group and 149 in the non-MACCE group)
were enrolled in Guangzhou City, Guangdong Province, China, and the samples
were randomly divided into a training set (n=372) and a validation set (n=159)
according to a 7:3 ratio. The variables were first screened by one-way logistic
regression analysis, then the modeling factors were determined by LASSO
regression, and finally the independent risk factors for MACCE were screened by
multifactorial logistic regression analysis, and the columnar graph model was
constructed. The performance of the model was evaluated by calibration curves,
internal bootstrap validation, ROC curve analysis, and validation set. Results:
One-way analysis of the training set showed statistically significant results
for balanced constitution, qi-deficient constitution, history of diabetes,
history of hypertension, elevated triglycerides, and elevated low-density
lipoprotein (LDL) (P<0.05). 15 predictors were identified by LASSO
regression, and 7 independent factors (qi-deficiency constitution,
phlegm-dampness constitution, history of diabetes, history of hypertension,
elevated triglycerides and LDL) were identified by multivariate analysis for
the construction of the columnar graph model. The calibration curve showed that
the model was well calibrated, with a training set C-index of 0.846 and a ROC
value of 0.835, and a validation set C-index of 0.857 and an AUC value of
0.844. Conclusion: Qi-deficiency constitution, phlegm-dampness constitution,
history of diabetes, history of hypertension, triglyceride elevation, and low-density
lipoprotein elevation are the independent risk factors for the occurrence of
cardiovascular and cerebrovascular events. The column-line graph model
developed in this study had good predictive efficacy and discrimination to
predict the risk of new cardiovascular and cerebrovascular events.
Key words：new-onset cardiovascular and cerebrovascular events; TCM constitution; indicators of physical examination; nomogram prediction model
发布时间：2026-01-06
点击量：49