中文摘要：

目的提高食品安全风险评估精度，构建化学污染物慢性膳食暴露评估全概率模型。方法利用我国膳食调查、污染物监测数据以及相应的人口学资料构建膳食暴露评估全概率模型。通过在消费量和污染物总体中进行蒙特卡洛（Monte Carlo）抽样，匹配相乘后得到暴露量的概率分布；构建贝塔二项正态分布（Betabinomial and normal，BBN）模型，将横断面调查获得的短期暴露量近似“拉伸”为长期（慢性）暴露量；通过Monte Carlo模拟和自助法（Bootstrap）对人群膳食暴露量进行变异度和不确定度分析。结果以江苏省居民铅膳食暴露评估为例，构建了化学污染物慢性膳食暴露全概率评估模型。全概率模型和半概率模型比较显示：两种模型计算的暴露量均值接近，但全概率模型计算结果的变异度大于半概率模型，表现为其低端百分位数小于半概率模型，高端百分位数大于半概率模型。结论化学污染物慢性膳食暴露全概率模型评估结果较半概率模型保守。

英文摘要：

Objective To establish a fully probabilistic model for evaluation of dietary exposure to chemical contami- nants and to improve the food safety risk assessment＇s accuracy. Methods Data from national diet and nutrition survey and food contamination monitoring program were used to establish fully probabilistic model. Monte Carlo was applied to conduct sampling from consumption and contamination data. Dietary exposure was then obtained through multiplying both data; Betabinomial and normal model was established to make the short-term exposure approximately represent the long-term exposure. Monte Carlo simulation and Bootstrap sampling were applied to analyze the variation and uncertainty. Results Based on the example of diet- ary exposure to Pb of Jiangsu population, fully probabilistic model for evaluation of dietary exposure to chemical contaminants was established. The mean dietary exposures calculated by two models were similar. Variation calculated by fully probabilistic model was higher than semi-probabilistic model. Compared with semi-probabilistic model, exposure of fully probabilistic model had higher high-end percentiles and lower low-end percentiles. Conclusion Fully probabilistic model are more conservative than semi-probabilistic model when assessing the long-term dietary exposure.