【摘要】 目的运用logistic逐步回归与ROC曲线分析对白血病中不同项目组合进行统计,建立急性白血病的初诊方程,以提高其诊断的准确性和特异性。方法本院初诊的白血病患者同步测定WBC、PLT、LDH、HGB,并统计每患者造血三系的变化(有一系异常则为VI,两系异常则为VII,三系异常则为VIII),采用logistic逐步回归,筛选鉴别诊断急性血液病中的相关指标,对不同组合进行综合分析,通过ROC曲线分析临床性能,最后得相应的数学模型即回归方程。结果LDH、WBC、PLT、VIII得出的方程是:P=1/[1+e-(-3.451x+0.005x1+0.108x2-0.08x3+2.016x4)],其灵敏度为84.6%,特异度为92.3%。结论运用ROC曲线和logistic回归综合分析不同项目组合简单有效,大大提高了诊断的科学性、准确性。更多还原

【Abstract】 Objective To establish the preliminary diagnosis equation of leukemia by using logistic stepwise regression and ROC curve to analyze the different proceeding in acute leukemia statistically,and improve the accuracy and specificity of diagnosis.Methods WBC,PLT,LDH and HGB were measured in the preliminary diagnosed acute leukemia patients,and the changes of three hemopoietic systems in all patients were determined(one abnormality for VI,two abnormalities for VII,three all changes for VIII).The correlative index of the differential diagnosis for acute hematonosis were sieved by logistic stepwise regression.Aggregate analysis were performed in different index combinations.Its clinic efficiency were analyzed by ROC curve.Based on above analysis,the mathematic model,i.e.regression equation,was established.Results The equation of LDH,WBC,PLT,VIII was P=1/[1+e-(-3.45x+0.005x1+0.108x2-0.08x3+2.016x4)],its sensibility was 84.6%,and specificity was 92.3%.Conclusions It was simple and utility by using ROC curve and logistic regression to aggregate analysis different index combinations,and can improve the accuracy in the diagnosis of acute leukemia.更多还原