Small sample robust approach to outliers and correlation of atmospheric pollution and health effects in Santiago de Chile

Adverse effects of air pollution on health are a global problem. Chile is no exception due to the increase of urban population and increasing pollution sources. For several years in the winter months in Santiago de Chile, environmental pre-emergency is decreed, which is due to the increase of measurements of contaminants and the risk that this means to health. In order to model the effects of pollution on health we consider a hierarchical Bayesian generalized linear mixed autoregressive model proposed by Ref. [18]. In particular, we apply the model to the number of children with respiratory diseases in the town of Santiago for the period June–August 2011, using the PM_2.5 data as covariate obtained by a spatiotemporal pollution model. In order to detect anomalous data, we apply to residuals both robust normality tests together with novel method of probabilities for mild or extreme outliers. We detected significant heterogeneity between stations which offer us better monitoring planning for the future. AMS 2010 subject classifications. Primary 60K10; Secondary 60E15, 60E05, 62F10.