Background: Various methods have been proposed in modelling spatio-temporal pattern of diseases in recent years. The construction of the spatio-temporal models can be either descriptive methods or dynamic. While the former mainly explores correlations by functions, the latter also depicts the process of transmission quantitatively. In this study, we aim to evaluate the differences in model fitting between a descriptive, spatio-temporal model and dynamic spatio-temporal model of schistosomiasis transmission in Guichi, Anhui Province, China.
Methods: The parasitological and environmental data at the village level from 1991 to 2014 were obtained by cross-sectional survey. The space-time changes of schistosomiasis risk were explored by two different spatio-temporal models, the Fixed Rank Kriging (FRK) model and the Integral-Difference Equation (IDE) model, and the performance of the two models in fitting schistosomiasis risk were compared.
Results: In both models, the average daily precipitation and the normalized difference vegetation index (NDVI) are significantly positively associated with schistosomiasis prevalence while distance to water bodies, the hours of daylight and day land surface temperature (LSTday) are significantly negatively associated. The overall root mean squared prediction error (MSPE) of the Integral-Difference Equation (IDE) and the FRK model were 0.35e-02 and 0.54e-02, respectively, and the correlation between the predicted and observed values of the IDE model (0.71) (p<0.01) was larger than the FRK one (0.53) (p=0.02). Our results also showed that the prediction error of the IDE model is lower than that of FRK model.
Conclusions: Regions close to rivers are key areas for further implementation of schistosomiasis prevention and control strategies. The IDE model fits better than the descriptive FRK model in capturing the geographic variation of schistosomiasis. Dynamic Spatio-temporal models have the advantage of quantifying the process of disease transmission and may provide more accurate predictions, which is of great importance with reference to future modelling.