Denoising a model employing automated bandwidth selection procedures and pre-whitened Euclidean-based quadratic surrogates in PROC ARIMA for optimizing asymptotic expansions and simulations of onchocerciasis endemic transmission zones in Burkina Faso

In this research we constructed multiple predictive ArcGIS Euclidean distance–based autoregressive infectious disease transmission oriented models for predicting geographic locations of endemic onchocerciasis (“river blindness”) transmission risk zones in Burkina Faso. We employed multiple spatiotemporal-sampled empirical ecological data sets of georeferenced covariates of riverine larval habitats of Similium damnosum s.l., a black fly vector of onchocerciasis and their surrounding villages with their retrospective tabulated prevalence rates. The estimators were regressed employing the modified sum of squares technique. The model also revealed that 5 to 10 km was mesoendemic, 10 to 15 was hypoendemic and after 15 km there was no transmission. Semi-parametric spatial filtering matrices, orthogonal eigenvectors and interpolated endmember signatures can be used to render robust ARIMA risk model residual forecasts by reducing latent unobservable error coefficients in regressed spatiotemporal field-sampled immature S. damnosum s.l. density count data for optimizing risk mapping of seasonal onchocerciasis endemic transmission zones.   Key words: Autoregressive integrated moving average (ARIMA), QuickBird, Similium damnosum s.l., onchocerciasis, Burkina Faso.