早期类风湿关节炎中采用ROC曲线评价各指标Logistic回归模型的应用价值

Application value of ROC curve in early rheumatoid arthritis for evaluation of Logistic regression models

目的探讨早期类风湿关节炎患者予以ROC曲线评价Logistic回归联合指标的应用价值。方法选取2017年9月~2018年8月我院收治的136例类风湿关节炎患者及38例同期健康体检者,检测其抗环瓜氨酸肽抗体(CCP)、C反应蛋白(CRP)、类风湿因子(RF)、红细胞沉降率(ESR),绘制ROC曲线并计算曲线下面积(AUC),把4个指标通过Logistic回归模型建立A组(CCP+RF)、B组(CCP+RF+CRP)、C组(CCP+RF+CRP+ESR),并分别进行比较。结果C组具有最大的AUC(P<0.05),A、B、C组的AUC分别为0.887、0.892、0.916;B组的特异性(94.96%)高于A组(91.77%)与C组(93.72%)(P<0.05);C组的灵敏度(86.54%)高于A组(79.54%)与B组(82.28%),差异有统计学意义(P<0.05)。结论通过ROC曲线评价Logistic回归模型得出,4项指标联合检测能够提高早期类风湿关节炎的诊断效能,为临床对类风湿关节早发现早治疗提供重要依据。

Objective To investigate the application value of receiver operating characteristic(ROC)curve evaluation in Logistic regression models in early rheumatoid arthritis patients.Methods A total of 136 patients with rheumatoid arthritis and 38 healthy subjects who were admitted to our hospital from September 2017 to August 2018 were enrolled in this study.The anti-cyclic citrullinated peptide antibody(CCP)and C-reactive protein(CRP),rheumatoid factor(RF),and erythrocyte sedimentation rate(ESR)were detected and ROC was plotted and area under ROC curve(AUC)was calculated.Using logistic regression,three groups were established based on the above-mentioned four indicators:group A(CCP+RF)and group B(CCP+RF+CRP),and group C(CCP+RF+CRP+ESR),and each of which was compared.Results The largest AUC in the group C,followed by group B,and group A in 0.916,0.892,and 0.887 accordingly(P<0.05).The specificity of group B was 94.96%,higher than that of group A(91.77%)and group C(93.72%)(P<0.05).The sensitivity of group C was 86.54%,higher than that of group A(79.54%)and group B(82.28%),with statistical differences(P<0.05).Conclusion The ROC curve evaluation on Logistic regression model shows that the joint detection of the four indicators can improve the diagnostic efficiency of early rheumatoid arthritis,and provide an important basis for clinical early detection of rheumatoid joints.