A model-based approach for evaluating exposure to extreme events in public health. A case study on the Lazio region

A new Bayesian model-based approach is introduced to deal with extreme epidemiological events in the Lazio region. Given that extreme temperatures and high pollutant concentrations harm the population, the aim is to develop an innovative method to assess exposure. The proposal leverages the Bayesian approach, potentially providing public decision-makers with tools to allocate health resources based on health risks. It begins by exploring data on extreme temperatures and pollutants (particulate 10 and 2.5, NO2, ozone), considering the geographical features of the region and predictors from the literature. A new modeling proposal of another study is then used, focused on quantile auto-regression models for extreme events. The main output of this stage will be exposure surfaces based on marginal quantiles varying over time. Next, a second-level Bayesian model is implemented to scale to the national level. Among various options, nearest neighbors Gaussian process techniques offer optimal computational and probabilistic properties. The final output will provide tools for assessing exposures in environmental epidemiology, aiding public health applications. Extreme temperature stress will be more relevant among older age groups, while extreme pollution events could be linked to diseases in young people, possibly connecting individual histories to exposure evolution.