Disease Mapping

Although Miguel A. Martinez-Beneito’s background is mostly based in mathematics/statistics his scientific career has been completely linked to Public Health. His first professional job was as statistician in the epidemiology unit of the Valencian regional health authority and all his research from then has been focused on the development of statistical methods for epidemiological studies. His main line of research is disease mapping and its extension to complex settings (multivariate spatial models, spatio-temporal models, spatial survival models, …) where he has published most of his research papers with either methodological/statistical or applied/epidemiological content. Regardless his peer-reviewed scientific publication Dr. Martinez-Beneito has been involved in several projects entailing the intensive application of disease mapping methods to the study of mortality in different contexts and regions. As a result he is author of 3 spatial atlas of mortality (2 of them corresponding to the Valencian region and another one to big Spanish cities) and 1 spatio-temporal atlas (http://www.geeitema.org/AtlasET/index.jsp?idioma=I). This extensive experience in geographical mortality studies makes Dr. Martinez-Beneito particularly suited to undertake this project. Paloma Botella-Rocamora’s background is based in mathematics/statistics, but her scientific career is mainly linked to statistics within Public Health. Her first scientific job was as part time research internship at the Epidemiology Unit of the Valencian regional health authority working in a project studying rare diseases, where she developed different spatial atlases of morbidity for rare diseases. During those years she also participated in the development of a spatial mortality atlas in the Valencian region, and a spatio-temporal mortality atlas in this same region (http://www.geeitema.org/AtlasET/index.jsp?idioma=I). She has also been the first author of the Spanish spatial atlas of rare diseases. She shared those jobs with her classes as part time associate professor at the University of Valencia and CEU-Cardenal Herrera University, where she already continues working as full time professor. Her teaching scope has always been linked to biostatistics in Health Sciences. Following the topic of his doctoral thesis, Paloma Botella Rocamora’s main line of research is disease mapping where she has published most of her research papers with either methodological/statistical or applied content. She has started to work in her recent scientific stay at the University of Minnesota (2013 summer) in the extension of disease mapping models to complex settings (multivariate spatial models, spatio-temporal models, …). This extensive experience in geographical mortality studies makes Dr. Botella-Rocamora particularly suited to undertake this project

I. DISEASE MAPPING: THE FOUNDATIONS



1. Introduction



Some considerations on this book



Notation



2. Some basic ideas of Bayesian inference



Bayesian inference



Some useful probability distributions



Bayesian Hierarchical Models



Markov chain Monte Carlo Computing



Convergence assessment of MCMC simulations



3. Some essential tools for the practice of Bayesian disease mapping



WinBUGS



The BUGS language



Running models in WinBUGS



Calling WinBUGS from R



INLA



INLA basics



Plotting maps in R



Some interesting resources in R for disease mapping practitioners



4. Disease mapping from foundations



Why disease mapping?



Risk measures in epidemiology



Risk measures as statistical estimators



Disease mapping, the statistical problem



Non-spatial smoothing



Spatial smoothing



Spatial distributions



The Intrinsic CAR distribution



Some proper CAR distributions



Spatial hierarchical models



Prior choices in disease mapping models



Some computational issues on the BYM model



Some illustrative results on real data



II. DISEASE MAPPING: TOWARDS MULTIDIMENSIONAL MODELING



5. Ecological Regression



Ecological regression: a motivation



Ecological regression in practice



Some issues to take care of in ecological regression studies



Confounding



Fallacies in ecological regression



The Texas sharpshooter fallacy



The ecological fallacy



Some particular applications of ecological regression



Spatially varying coefficients models



Point source modelling



6. Alternative spatial structures



CAR-based spatial structures



Geostatistical modeling



Moving-average based spatial dependence



Splines based modeling



Modelling of specific features in disease mapping studies



Modeling partitions and discontinuities



Models for fitting zero excesses



7. Spatio-temporal disease mapping



Some general issues in spatio-temporal modelling



Parametric temporal modelling



Splines-based modelling



Non-parametric temporal modelling



8. Multivariate modelling



Conditionally specified models



Multivariate models as sets of conditional multivariate Distributions



Multivariate models as sets of conditional univariate distributions



Coregionalization models



Factor models, Smoothed ANOVA and other approaches



Factor models



Smoothed ANOVA



Other approaches



9. Multidimensional modelling



A brief introduction and review of multidimensional modeling



A formal framework for multidimensional modeling



Some tools and notation



Separable modeling



Inseparable modeling



Annex 1



Bibliography



Index