摘要
目的:探索诊断试验的灵敏度与特异度、似然比、预测值的变化关系。方法:应用数学方法证明了确定人群中提高灵敏度(Se)时,特异度(Sp)、阳性似然比(+LR)、阴性似然比(-LR)、阳性预测值(+PV)、阴性预测值(-PV)的数学变化关系,并给出了实例分析。结果:提高灵敏度时,特异度、预测值和似然比的升高或降低,与病人诊断试验阳性数的增量、非病人诊断试验阳性增量等有关。结论:提高灵敏度时,特异度降低或不变,似然比和预测值存在升高、不变或降低等情况;阴性似然比与阴性预测值的变化方向相反。
Objective. To explore mathematical relation of sensitivity and specificity, likelihood ratio, predictive value in diagnostic test. Methods： The change of specificity, likelihood ratio and predictive value caused by increasing sensitivity was proved by mathematics, and two examples were analyzed. Results. Increasing sensitivity, change of specificity, likelihood ratio and predictive value was associated with increments （ Aa and Ab ） of positive cases diagnosed by test in patients and non-patients. Conclusion： When sensitivity was increased, specificity could be not changed or reduced, likelihood ratio and predictive value could be not changed, or increased, or reduced. --LR Changes in the opposite direction to --PV.
出处
《数理医药学杂志》
2008年第5期574-576,共3页
Journal of Mathematical Medicine
关键词
诊断试验
灵敏度
特异度
似然比
预测值
diagnostic test
sensitivity
specificity
likelihood ratio
predictive value
作者简介
许汝福（1963-），男，四川成都人，副教授。研究方向：统计方法与统计预测、传染病流行病学等。
