Spatiotemporal filtering modeling of hand, foot, and mouth disease: a case study from East China, 2009–2015

Hand, foot, and mouth disease (HFMD) shows spatiotemporal heterogeneity in China. A spatiotemporal filtering model was constructed and applied to HFMD data to explore the underlying spatiotemporal structure of the disease and determine the impact of different spatiotemporal weight matrices on the results. HFMD cases and covariate data in East China were collected between 2009 and 2015. The different spatiotemporal weight matrices formed by Rook, K-nearest neighbour (KNN; <i>K</i> = 1), distance, and second-order spatial weight matrices (<i>SO-SWM</i>) with first-order temporal weight matrices in contemporaneous and lagged forms were decomposed, and spatiotemporal filtering model was constructed by selecting eigenvectors according to <i>MC</i> and the <i>AIC.</i> We used <i>MI</i>, standard deviation of the regression coefficients, and five indices (<i>AIC</i>, <i>BIC</i>, <i>DIC</i>, <i>R</i>², and <i>MSE</i>) to compare the spatiotemporal filtering model with a Bayesian spatiotemporal model. The eigenvectors effectively removed spatial correlation in the model residuals (Moran's <i>I</i> < 0.2, <i>p</i> > 0.05). The Bayesian spatiotemporal model's Rook weight matrix outperformed others. The spatiotemporal filtering model with <i>SO-SWM</i> was superior, as shown by lower <i>AIC</i> (92,029.60), <i>BIC</i> (92,681.20), and <i>MSE</i> (418,022.7) values, and higher <i>R</i>² (0.56) value. All spatiotemporal contemporaneous structures outperformed the lagged structures. Additionally, eigenvector maps from the Rook and <i>SO-SWM</i> closely resembled incidence patterns of HFMD.