Spatiotemporal varying coefficient model for respiratory disease mapping in Taiwan.

Respiratory diseases have been global public health problems for a long time. In recent years, air pollutants as important risk factors have drawn lots of attention. In this study, we investigate the influence of $\pm2.5$ (particulate matters in diameter less than 2.5 ${\rm{\mu }} m$) on hospital visit rates for respiratory diseases in Taiwan. To reveal the spatiotemporal pattern of data, we propose a Bayesian disease mapping model with spatially varying coefficients and a parametric temporal trend. Model fitting is conducted using the integrated nested Laplace approximation, which is a widely applied technique for large-scale data sets due to its high computational efficiency. The finite sample performance of the proposed method is studied through a series of simulations. As demonstrated by simulations, the proposed model can improve both the parameter estimation performance and the prediction performance. We apply the proposed model on the respiratory disease data in 328 third-level administrative regions in Taiwan and find significant associations between hospital visit rates and $\pm2.5$.