chapter  23
Bayesian Variable Selection in Semiparametric and Nonstationary Geostatistical Models: An Application to Mapping Malaria Risk in Mali
ByFederica Giardina, Nafomon Sogoba, Penelope Vounatsou
Pages 18

Geostatistical models are used to analyze data collected at a discrete set of locations (georeferenced data) within a continuous domain (see Chapter 11 in this volume). They have been widely applied to problems ranging from geology and ecology to epidemiology and public health (Gelfand et al., 2004). Applications in epidemiology are mainly concerned with relating disease data to a set of predictors (i.e., environmental or climatic variables)

T&F Cat #K23899 — K23899 C023 — page 399 — 2/20/2016 — 12:19

T&F Cat #K23899 — K23899 C023 — page 400 — 2/20/2016 — 12:19


with the aim of determining the main risk factors and predicting disease outcome measures (e.g., risk, incidence, and mortality) at unobserved locations (Lawson, 2013). The Bayesian formulation of linear and generalized linear geostatistical models has been introduced by Diggle et al. (1998).