Bayesian hierarchical model-based food pollutant exposure evaluation method

Abstract

The invention relates to a food pollutant exposure assessment method based on a Bayesian hierarchical model, which optimizes the calculation of the average concentration of pollutants in food, adopts the Bayesian hierarchical model to calculate and restore the known data in the literature, derives the concentration distribution probability, and then carries out the calculation of exposure assessment. According to the method, the total data, such as average value data and variance data, obtained in literatures can be summarized, random sample points of posteriori distribution are extracted by combining prior distribution, using a Bayesian hierarchical model and adopting an MCMC sampling method, so that concentration distribution of detected pollutants is obtained, and then exposure evaluation is calculated by combining consumption data. According to the method, the data can be corrected, and the data can be restored more accurately.

Description

technical field [0001] The invention relates to the field of dietary heavy metal exposure assessment methods, in particular to a food pollutant exposure assessment method based on a Bayesian hierarchical model. Background technique [0002] The exposure assessment of food pollutants is to combine the data of the concentration of pollutants in food and the data of food consumption, and use statistical methods to calculate the exposure of pollutants in food. Food exposure models can be divided into the following four categories according to their principles: (1) deterministic models based on point estimation; (2) probability assessment models based on empirical estimation; (3) probability assessment models based on parameter estimation; (4) probability assessment models based on Probabilistic evaluation models for Bayesian statistics. Traditional dietary exposure assessment methods rely on sample data and do not involve the use of prior information. Bayesian statistics combi...

AI Technical Summary

Problems solved by technology

Bayesian statistics combines sample data with prior information, and adjusts the sample distribution on the basis of prior information. Therefore, Bayesian statistics does not rely too much on sample information, and accurate prior information can make up for the lack of sample data. Insufficient, to solve the following problems in the traditional model: (1) The sample size is too small: In this case, the difference between the data generated by repeated sampling is small, especially at the tail end of the distribution, and the traditional probability assessment method is often limited

(2) No original data: When there is only summary data such as mean and standard deviation without original data, the evaluation effect of traditional methods is often poor

(3) There is a hierarchical structure in the data: for example, the hierarchical structure composed of regions, households, individuals, and consumption days. Traditional probability models are often difficult to deal with such responsible hierarchical data organizations

Images