Thinking outside the curve, part I modeling birthweight distribution 英文参考文献.docVIP

  • 8
  • 0
  • 约4.24万字
  • 约 12页
  • 2017-05-16 发布于上海
  • 举报

Thinking outside the curve, part I modeling birthweight distribution 英文参考文献.doc

Thinking outside the curve, part I modeling birthweight distribution 英文参考文献

Charnigoetal.BMCPregnancyandChildbirth2010,10:37 /1471-2393/10/37 TECHNICAL ADVANCE OpenAccess Thinkingoutsidethecurve,partI: modelingbirthweightdistribution RichardCharnigo1*,LorieWChesnut2,TonyLoBianco3,RussellSKirby4 Abstract Background:Greaterepidemiologicunderstandingoftherelationshipsamongfetal-infantmortalityandits prognosticfactors,includingbirthweight,couldhavevastpublichealthimplications.Akeysteptowardthat understandingisarealisticandtractableframeworkforanalyzingbirthweightdistributionsandfetal-infant mortality.Thepresentpaperisthefirstofatwo-partseriesthatintroducessuchaframework. Methods:Weproposedescribingabirthweightdistributionviaanormalmixturemodelinwhichthenumberof componentsisdeterminedfromthedatausingamodelselectioncriterionratherthanfixedapriori. Results:Weaddressanumberofmethodologicalissues,includinghowthenumberofcomponentsselected dependsonthesamplesize,howthechoiceofmodelselectioncriterioninfluencestheresults,andhowestimates ofmixturemodelparametersbasedonmultiplesamplesfromthesamepopulationcanbecombinedtoproduce confidenceintervals.Asanillustration,wefindthata4-componentnormalmixturemodelreasonablydescribesthe birthweightdistributionforapopulationofwhitesingletoninfantsborntoheavilysmokingmothers.Wealso comparethis4-componentnormalmixturemodeltotwocompetitorsfromtheexistingliterature:acontaminated normalmodelanda2-componentnormalmixturemodel.Inasecondillustration,wediscoverthata6-component normalmixturemodelmaybemoreappropriatethana4-componentnormalmixturemodelforageneral populationofblacksingletons. Conclusions:Theframeworkdevelopedinthispaperavoidsassumingtheexistenceofanintervalofbirthweights overwhichtherearenocompromisedpregnanciesanddoesnotconstrainbirthweightswithincompromised pregnanciestobenormallydistributed.Thus,thepresentframeworkcanrevealheterogeneityinbirthweightthat isundetectableviaacontaminatednormalmodelora2-componentnormalmixturemodel. Background epidemiologicunderstandingoftherelationshipsamong The impact of birthweig

您可能关注的文档

文档评论(0)

1亿VIP精品文档

相关文档