Methods to investigate the effects of multiple air pollution constituents

We are focusing on several methodological aspects related to the link between multiple metrics of air pollution and health.

  • Source apportionment of ambient particle matter, which uses Bayesian nonparametric processes with dependence on dynamic factors (e.g. wind speed and direction) to model the underlying spatial or temporal structure and the distribution of contaminants to identify sources. We then assess the link between the apportioned sources and health outcomes in order to move from the standard epidemiological models where total PM is considered to a source-contribution evaluation, which is more relevant for decision making.
  • Two-component Bayesian hierarchical model for multiple air pollutants, where the pollutant model to estimate the “true” concentrations is linked to the health outcome in a time-series perspective.
Marta Blangiardo
Marta Blangiardo
Professor of Biostatistics

Related