结合Kaplan-Meier法与Bootstrap法处理膳食暴露评估中的未检出值

Dealing with non-detects for dietary exposure assessment using Kaplan-Meier methods and Bootstrap

目的:探讨应用Kaplan-Meier方法与bootstrap方法相结合处理膳食暴露评估中多个检出限存在时的未检出值数据.方法:以大米中的镉残留数据为例,应用Kaplan-Meier方法对数据拟合累积概率分布曲线,并且应用bootstrap方法估计残留值的均数和中位数的标准误以及累积概率的区间.结果:Kaplan-Meier方法处理多个检出限存在时的未检出值简单,便于计算机实现,Bootstrap估计的区间与近似估计区间一致性好.结论:对环境暴露评估中包含未检出值的左截尾数据的处理,结合bootstrap区间估计的Kaplan-Meier方法是值得推荐的有效方法.

AIM：To combine Kaplan-meier method and Bootstrap estimation in dealing with the left-censored data for exposure assessment. METHODS：The cumulative probability distribution for the cadmium residue concentration in rice was fitted by Kaplan-Meier method,basing on which standard errors and interval estimates of mean,median et al. were gained by bootstrap method. RESULTS：The Kaplan-Meier method can be used simply and conveniently by software. The intervals estimated by bootstrap were consistent with those by approximate method.CONCLUSION： Kaplan-Meier method combined with bootstrap is efficient method worth to recommend in dealing with data including left-censored values.