多种污染物混合暴露对健康影响统计模型的研究进展  

Research progress of statistical models on the health effects of mixed exposure to multiple pollutants

作　　者：侯静 张莉 颜豪森 李佳琏 郭佳珂 姚竞帆 孙红卫 HOU Jing;ZHANG Li;YAN Hao-sen;LI Jia-lian;GUO Jia-ke;YAO Jing-fan;SUN Hong-wei(School of Public Health and Management,Binzhou Medical University,Yantai 264003,China)

机构地区：[1]滨州医学院公共卫生与管理学院,山东烟台264003

出　　处：《预防医学论坛》2023年第10期794-800,共7页Preventive Medicine Tribune

基　　金：国家级大学生创新训练项目(202210440057)。

摘　　要：近年来,环境中化学混合物的不良健康效应备受关注。但化学混合物成分的高维性和高度相关性使得传统的统计方法存在共线性和方差膨胀问题。在多污染物模型中,评估环境因素对健康影响的挑战包括但不限于确定污染物混合物中最关键的成分、检查潜在的相互影响以及多污染物共同暴露引起人体不良健康影响的整体效应。本文综述了用于评估多污染物混合暴露对健康影响的统计方法,如累积风险指数、加权分位数和回归、分位数G计算、主成分分析以及监督主成分分析、最小绝对收缩和选择算子、Group-LASSO交互网模型、D/S/A算法、贝叶斯核机器回归等,系统总结每种方法的应用优势、不足和结果可解释性等,为研究者选择多污染物混合对健康影响的统计模型提供参考依据。In recent years,there has been significant attention given to the adverse health effects caused by the presence of chemical mixtures in the environment.Traditional statistical methods encounter challenges in dealing with covariance and variance inflation due to the high-dimensionality and high correlation among the components of chemical mixtures.When assessing the health effects of environmental factors using multi-pollutant modelling,several challenges arise.These challenges include,but are not limited to,identifying the most critical components within the pollutant mixture,examining potential interactions,and evaluating the overall impact of co-exposure to multiple pollutants on inducing adverse health effects in humans.In this study,we investigate various statistical methods commonly used to assess the health effects of mixed multi-pollutant exposures.These methods include Cumulative risk index(CRI),Weighted quantile sum(WQS),Quantile G calculation,Principal component analysis(PCA),Supervised principal component analysis(SPCA),Least absolute shrinkage and selection operator(LASSO),Group-LASSO interaction network model,D/S/A algorithm,Bayesian kernel machine regression(BKMR),and more.We systematically summarize the advantages,limitations,and interpretability of each method,aiming to provide researchers with a reference for selecting appropriate statistical models when studying the health impacts of multi-pollutant mixing.

关 键 词：污染物 混合暴露 统计模型 健康效应 

分 类 号：R821.31[医药卫生—临床医学] R12