Abstract
Resumo: Durante muito tempo o consumo de produtos fumados e vinho tinto era apontado como o principal factor de risco do cancro de estômago. No entanto, tem-se sugerido mais recentemente que a existência de uma bactéria, chamada Helicobacter Pylori, no interior do estômago de uma grande percentagem da população pode ser a maior responsável por esta doença tão mortal. A falta de dados relativos à prevalência desta bactéria na população em risco, em Portugal, leva-nos a ter de, no âmbito de uma modelação espacial hierárquica Bayesiana, utilizar variáveis substitutas e, consequentemente, a ter de adaptar modelos de forma a que estes acomodem esta situação. Apresentam-se e aplicam-se duas abordagens com diferentes fundamentações metodológicas a este problema, genericamente designado por uso de covariáveis com erros de medição, justificando-se os distintos resultados obtidos.
Abstract: For a long time smoked food and red wine consumption were pointed as the main risk factors for stomach cancer. More recently, however, it has been considered that the existence of a certain bacterium, called Helicobacter Pylori, in the stomachs of a huge percentage of population might be the great responsible for this deadly disease. The lack of data regarding the prevalence of this bacterium in theat risk population, in Portugal, has led us to use surrogate covariates, in a spatial hierarchical Bayesian framework, and consequence adaptation of models in order to accommodate this situation. For this problem, generically designated as the use of covariates with measurement errors, we present and apply here two approaches with different methodological fundamentals, justifying the different results that we obtain.
Abstract: For a long time smoked food and red wine consumption were pointed as the main risk factors for stomach cancer. More recently, however, it has been considered that the existence of a certain bacterium, called Helicobacter Pylori, in the stomachs of a huge percentage of population might be the great responsible for this deadly disease. The lack of data regarding the prevalence of this bacterium in theat risk population, in Portugal, has led us to use surrogate covariates, in a spatial hierarchical Bayesian framework, and consequence adaptation of models in order to accommodate this situation. For this problem, generically designated as the use of covariates with measurement errors, we present and apply here two approaches with different methodological fundamentals, justifying the different results that we obtain.
Original language | Portuguese |
---|---|
Title of host publication | Estatística jubilar |
Subtitle of host publication | Actas do XII congresso Anual da Sociedade Portuguesa de Estatística |
Editors | Carlos Braumann, Paulo Infante, Maria M. Oliveira, Russell Alpízar-Jara, Fernando Rosado |
Place of Publication | Lisboa |
Publisher | Sociedade Portuguesa de Estatística |
Pages | 477-488 |
Number of pages | 12 |
ISBN (Print) | 972-8890-04-4 |
Publication status | Published - 2005 |
Event | XII Congresso Anual da Sociedade Portuguesa de Estatística - Évora, Portugal Duration: 29 Sep 2004 → 2 Oct 2004 |
Conference
Conference | XII Congresso Anual da Sociedade Portuguesa de Estatística |
---|---|
Country/Territory | Portugal |
City | Évora |
Period | 29/09/04 → 2/10/04 |
Keywords
- Modelação espacial
- modelos hierárquicos
- Bayesianos
- modelos com erros nas covariáveis
- modelação conjunta de doenças
- Spatial modelling
- Hierarchical Bayesian models
- models with errors in covariates
- joint modelling of diseases.