Estimating short-term air pollution effects on health via spectral methods

Environmental epidemiology estimation of air pollution effects on health is often conducted via generalized additive mixed models and observa- tional data aggregated over space and time. Often times, the introduction of structured random effects can produce substantial changes in the slope coeffi- cient which makes it challenging to interpret effect size. Spectral methods to model the slope as a function of spatial scale have recently been introduced by Guan et al (2022). We extend those methods to the spatio-temporal setting motivated by a case study on estimation of short-term air pollution effects on acute hospital admissions.