目的 探索支持向量机方法在建立冠心病早期诊断模型中的应用,为冠心病危险因素在早期诊断中的合理应用提供理论依据.方法 首先应用logistic回归分析方法筛选冠心病危险因素,将有统计学意义的危险因素与24 h动态心电图检查结果共同构建支持向量机模型,并应用测试数据集对各模型的诊断能力进行评价.结果 24 h动态心电图检查结果与危险因素共同构建的支持向量机模型较单独应用24 h动态心电图诊断有更好的诊断准确率和灵敏度,特异度较低.对应用不同变量构建的模型进行比较,应用24 h动态心电图,结合年龄、性别、糖尿病、高血压构建的模型诊断效果较好,准确率为70.35%,灵敏度为90.27%,特异度为34.76%.结论 应用支持向量机可以建立合适的冠心病早期诊断模型;结合主要危险因素进行冠心病的早期诊断可以提高诊断准确率.%Objective To explore the application of the Support Vector Machine(SVM) in the diagnosis of Coronary Artery Disease (CAD) ;And to provide the theory basis for the usage of risk factors in the early diagnosis. Methods Backward logistic regression was used to choose significant variables. We used significant variables and 24-honr holter to build the SVM. Then different models were evaluated with the same test dataset. Results The accuracy and sensitivity of the SVM which was built with risk factors were higher than 24-hour holter to diagnose CHD,and the specificity was lower. After contrasting the diagnostic capabilities among different SVM models, we found that the model built with 24-hour holter,combined with age,sex,diabetes,hypertension was better. The accuracy was 70.35％,the sensitivity was 90.27％ and the specificity was 34. 76％. Conclusion SVM could be used as the early diagnostic method for CHD,and the accuracy of early diagnosis would be higher in consideration of major risk factors.
展开▼
