摘要
目的:探讨logistic多指标联合诊断试验ROC分析中的应用,评价4种与冠心病发病有关指标在冠心病诊断及联合诊断中的效果。方法:根据疾病状态建立logistic回归模型,通过形成的预测概率或联合预测因子为分析指标,并结合双正态模型建立ROC曲线。结果:通过实例阐述了整个分析过程,并说明了指标对诊断冠心病的有效性,确定了基于联合诊断的最佳工作点。结论:ROC分析中结合logistic回归模型简单有效,尤其适用于有协变量或多指标联合诊断试验的分析评价。
Objective： To explore the application of logistic model in ROC curve analysis, and to evaluate the clinical value and discriminatory power of 4 biornarkers in diagnosing Coronary Heart Disease and determine the Optimal Operating Point （OOP） by using the Receiver Operating Characteristic （ROC） curves. Method： Based on the logistic model, combining predictors or prohahil ities were gained and applied to establish empirical and binomial model ROC curves, Results： An example for prediction of coronary heart disease was presented to illustrate the whole analysis steps. At the same time, the method shows that the HDL and TG were suitable for coronary heart disease screening, and the optimal operating point was determined for the combining screening test. Conclusion： Roc analysis based on the logistic model is easy and convenient, especially using in the screening test with covariates or multiple markers for classification.
出处
《现代预防医学》
CAS
北大核心
2006年第5期723-724,740,共3页
Modern Preventive Medicine
