資源描述:
《中國(guó)膳食暴露評(píng)估非參數(shù)概率模型構(gòu)建》由會(huì)員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在工程資料-天天文庫。
1、萬方數(shù)據(jù)生堡亟隨醫(yī)堂盤查墊!璺生!旦筮竺鲞筮!期g墮!!墮叢鯉:叢型!墊!Q:叢竺:№三中國(guó)膳食暴露評(píng)估非參數(shù)概率模型構(gòu)建孫金芳劉沛陳炳為陳啟光余小金王燦楠李靖欣【摘要】目的為提高評(píng)估精度并與國(guó)際食品安全風(fēng)險(xiǎn)評(píng)估技術(shù)接軌,構(gòu)建中國(guó)膳食暴露評(píng)估非參數(shù)概率模型。方法利用我國(guó)膳食調(diào)查、污染物監(jiān)測(cè)數(shù)據(jù)及相應(yīng)的人口學(xué)資料建立膳食消費(fèi)量和化學(xué)污染物濃度經(jīng)驗(yàn)分布。通過蒙特卡洛(MonteCarlo)模擬和自助法(Bootstrap)抽樣獲得人群膳食暴露變異度和不確定度。其中,膳食量數(shù)據(jù)和人口學(xué)數(shù)據(jù)來源于2002年中國(guó)居民營(yíng)養(yǎng)與健康狀況調(diào)查24h膳食回顧法收集的22567個(gè)家庭
2、66172人連續(xù)3d調(diào)查共計(jì)193814個(gè)人目、1810703條數(shù)據(jù)。污染物監(jiān)測(cè)數(shù)據(jù)為2000--2006年全國(guó)14個(gè)省或地區(qū)食品污染物監(jiān)測(cè)網(wǎng)以及2005--2006年海關(guān)出口農(nóng)產(chǎn)品監(jiān)測(cè)數(shù)據(jù),包括重金屬、農(nóng)藥,以及霉菌毒素(如黃曲霉毒素)等135種污染物,涉及499種食物,共計(jì)487819條數(shù)據(jù)。結(jié)果構(gòu)建了包括重金屬、農(nóng)藥及部分毒素的我國(guó)人群膳食暴露非參數(shù)概率評(píng)估模型,得到不同污染物膳食暴露量分布的指標(biāo)統(tǒng)計(jì)量和95%可信區(qū)間。對(duì)7一10歲兒童乙酰甲胺磷膳食暴露評(píng)估顯示,城鄉(xiāng)兒童膳食暴露量的中位數(shù)分別為1.77ttg·kg~·d“和2.48峙-l‘g~·d~,其9
3、5%可信區(qū)間分別為(1.59—2.06)懈·kg~·d。1和(2.33—2.80)ttg·kg~·d~。結(jié)論構(gòu)建的非參數(shù)概率模型可量化暴露評(píng)估中的變異度和不確定度,提高了膳食暴露評(píng)估精度?!娟P(guān)鍵詞】模型,統(tǒng)計(jì)學(xué);統(tǒng)計(jì)學(xué),非參數(shù);膳食調(diào)查;危險(xiǎn)性評(píng)估Establishmentofnon-parametricprobabilisticmodelforevaluationofChinesemetaryexposureSUNJin扣昭,UUPei,CHEN歷昭一wei,CHENQZ-gunng,YUXiao-jin,WANGCan-nan,LI.1ing·xin.Dep
4、artmentof眥miotograndBiostatistics,SchoolofPublicHealth,SoutheastUniversity,Nanjing210009,ChinaCorrespondingaut.110r:刪&i。Email:liupeiseu@126.com.【Abstract】ObjectiveToestablishanon—parametrieprababilistiemodelforevaluationofChinesedietaryexposureandtoimprovetheassessmentaccuracywhilein
5、tegratingintotheglobalriskassessmentonfoodsafety.MethodsContaminationdataWaSfromthenationalfoodcontaminationmonitoringprogramduring2000一2006,includingheavymetals,pesticidesandmycotoxins,amountingto135contaminantswith499commoditiesand487819samples.FoodconsumptiondataWasobtainedfromthe
6、nationaldietandnutritionsurveyconductedin2002withthreeconsecutivedaysby24一hourrecallmethod,and66172consumerswereincluded.MonteCarlosimulationWasappliedtoderivetheintakedistribution.a(chǎn)ndtheuncertaintyofeachpercentileWasestimatedusingtheBootstrapsampling,ResultsDifferentnon—parametricpr
7、obabilisticmodelsfordietaryexposureevaluationonheavymetals,pesticidesand$ouleofthetoxinswereestablishedforChinesepeople,andintakedistributionswith95%confidenceintervalsofthesecontaminantswereestimsted.TBkingacephateagallexample.theresultsofitsmodelshowstllat.forthe7—10year-oldchildre
8、n,tllemedian