论文部分内容阅读
目的应用决策树模型(CART)研究MSM人群HIV感染者总淋巴细胞计数(TLC)与CD_4~+T细胞计数的相关性,并与常规ROC分类方法比较。方法选取137例MSM人群HIV感染者,所有病例均未接受抗病毒治疗,采集203例血样标本,检测血常规指标和CD_4~+T细胞计数,把与CD_4~+T细胞计数显著相关的血常规指标纳入CART模型,计算灵敏度、特异度、阳性预测值(PPV)、阴性预测值(NPV)和约登指数。结果 203例血样标本的总淋巴细胞计数(TLC)与CD_4~+T细胞计数分别为(1 771.9±720.3)cells/μL、(410.9±201.4)cells/μL,两者呈显著正相关(r=0.603,P<0.001)。CART模型分类CD_4~+T细胞计数≤350 cells/μL的灵敏度、特异度、PPV、NPV和约登指数分别为66.7%、70.4%、58.4%、77.2%和0.371,CART模型分类CD_4~+T细胞计数≤500 cells/μL的灵敏度、特异度、PPV、NPV和约登指数分别为93.8%、53.5%、83.4%、77.5%和0.473;两个CART模型均优于ROC分类方法。结论 TLC与CD_4~+T细胞计数显著相关,TLC分类预测MSM人群HIV感染者CD_4~+T细胞计数具有良好的灵敏度和特异度,在资源有限地区可以用于MSM人群艾滋病疾病进展监测,应用CART模型纳入多个指标联合预测的效果更佳。
Objective To study the correlation between total lymphocyte count (TLC) and CD_4 ~ + T cell count in HIV-infected MSM population by using the decision tree model (CART) and to compare with conventional ROC classification. Methods Totally 137 cases of HIV-1 infected MSM population were selected. All cases were not treated with antiviral therapy. 203 blood samples were collected and blood routine indexes and CD_4 ~ + T cell counts were detected. Blood routinely correlated with CD_4 ~ + T cell counts Indicators were included in the CART model to calculate sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV) and Youden index. Results The total lymphocyte count (TLC) and CD_4 ~ + T cell counts of 203 blood samples were (1 771.9 ± 720.3) cells / μL and (410.9 ± 201.4) cells / μL, respectively, 0.603, P <0.001). The sensitivity, specificity, PPV, NPV and Youden index of CART model classification CD_4 ~ + T cell count ≤350 cells / μL were 66.7%, 70.4%, 58.4%, 77.2% and 0.371 respectively. The CART model classified CD_4 ~ + T cells The sensitivity, specificity, PPV, NPV and Youden index for counts ≤ 500 cells / μL were 93.8%, 53.5%, 83.4%, 77.5% and 0.473, respectively; both CART models outperformed the ROC classification. Conclusion TLC has significant correlation with CD_4 ~ + T cell count. TLC classification predicts the sensitivity and specificity of CD_4 ~ + T cell count in HIV infected MSM population. It can be used for monitoring the progress of HIV / AIDS in MSM population in areas with limited resources. The model incorporates multiple indicators of joint prediction better.