Analysis of heat waves effects on health using GAM and bootstrap based model selection.

It is known that high summer temperature may lead to worsening health conditions among fragile individuals within exposed populations. It is also argued that multi-day patterns of high temperature – heat waves – may have relevant effects on health. We will discuss the possible measures of heat waves intensity to be included in a generalized additive model explaining the number of hospital admissions occurred during summer months in Milano. The issue of variable selection is central to the analysis: a computational method is discussed which may help in assessing the robustness of model selection method. Eventually, we obtain evidence supporting the relevance of heat waves in driving adverse health episodes.