ROC 曲线和Logistic 回归评估PAF 联合NFATc1 对冠心病的诊断价值

ROC curve and logistic regression analysis in assessment of value of joint detection of PAF and NFATc1 in diagnose of coronary heart disease

摘要:

目的 探讨ROC 曲线和Logistic 回归评估血小板活化因子（PAF）和活化T 淋巴细胞核因子c1 （NAFTc1）对冠状动脉粥样硬化性心脏病（冠心病）的诊断价值。方法 选取2014 年1 月-2016 年1 月该院 就诊的188 例疑似冠心病患者为研究对象，根据冠状动脉造影结果分为对照组和冠心病组，其中对照组56 例， 冠心病组132 例。采用ELISA 法检测两组患者血清PAF、NAFTc1 浓度，采用Logistic 回归方程和ROC 曲线判 定单项检测和联合检测在冠心病诊断中的价值。结果 对照组和冠心病组血清PAF 浓度分别为（8.95±5.74）和 （18.13±4.72）ng/ml，NAFTc1 浓度分别为（22.78±3.15）和（30.58±5.24）pg/ml，两组比较差异有统计学意 义（P <0.05）。以PAF、NAFTc1 2 个单因素为自变量，进行二项分类Logistic 回归分析，得到Logistic 回归模 型：Logit（P）=-8.463+0.417PAF+0.098NAFTc1，该模型诊断准确率为85.64%。PAF、NAFTc1、Logistic 回 归模型联合检测的曲线下面积分别为0.806、0.815 及0.900，差异有统计学意义（P <0.05），Logistic 回归模型 联合检测的95%CI（0.838，0.962）。PAF 检测阈值为9.75 ng/ml，NAFTc1 检测阈值为23.95 pg/ml，Logistic 回归 模型联合检测敏感性为0.86，特异性为0.71。结论 PAF 联合NFATc1 检测效果优于单项检测，可用于冠心病筛查， 具有较高的诊断价值。

Abstract:

Objective To study the effect of ROC curve and logistic regression in the assessment of the value of joint detection of PAF and NFATc1 in diagnosis of coronary heart disease. Methods The study objects included 188 patients of suspected coronary heart disease in our hospital from January 2014 to January 2016. They were divided into a control group (56 cases) and a coronary heart disease group (132 cases) according to the results of coronary arterial angiography. ELISA was used to detect the serum PAF and NFATc1 concentrations in the two groups of patients. ROC curve and logistic regression equation were used to validate the detection value of single test and joint test. Results The serum PAF concentration was (8.95 ± 5.74) ng/ml in the control group and (18.13 ± 4.72) ng/ml in the coronary heart disease group, which showed a significant difference (P < 0.05). The serum NAFTc1 concentrations of the control group and the coronary heart disease group were (22.78 ± 3.15) pg/ml and (30.58 ± 5.24) pg/ml respectively, there was a significant difference (P < 0.05). When PAF and NAFTc1 were taken as independent variables to make two categories of logistic regression analysis, the logistic regression model was: Logit (P) = -8.463 + 0.417PAF + 0.098 NAFTc1, its diagnostic accuracy was 85.64%. The area under the curves of PAF, NAFTc1 and logistic regression joint detection was 0.806, 0.815 and 0.900 separately, which showed significant differences (P < 0.05), the 95% CI of the joint detection was (0.838, 0.962). The detection threshold of PAF and NAFTc1 was 9.75 ng/ml and 23.95 pg/ml respectively, the sensitivity of logistic regression joint detection was 0.86, the specificity was 0.71. Conclusions Joint detection of PAF and NFATc1 is better than single detection, can be applied in the screening of coronary heart disease, and has high diagnostic value.