Spatial prediction of flea index of transmitting plague based on environmental similarity. Issue 3 (2nd July 2020)

• Record Type: Journal Article
• Authors: Du, Haiwen; Zhu, A-Xing; Wang, Yong
• Is Part Of: Annals of GIS. Volume 26:Issue 3(2020)
• Page Start: 227
• Page End: 236
• Publication Date: 2020-07-02
• Subjects: Plague -- flea index -- geographic environmental variables -- spatial prediction -- third law of geography; Geographic information systems -- Periodicals; Periodicals; 910.285
• DOI: 10.1080/19475683.2020.1788639
• Languages: English
• ISSNs: 1947-5683

Abstract

ABSTRACT: Plague is a natural infectious disease which both people and mammals can contract once bitten by infected parasitic fleas. Flea index is an important outbreak indication. Outbreaks are more likely to occur for areas where flea index values are high. Thus, Prediction of spatial variation in flea index is vital to the prevention and control of the related plague. Existing methods of spatial prediction techniques require extensive field data to develop the needed models and limited for areas with insufficient field data. This paper presents an approach based on environmental similarity. The basic idea is that the more similar the environment of a location to locations at which the flea index is high the more likely that the flea index at this location is high as well. A method based on this idea was developed to predict spatial distribution of flea index of plague based on environmental similarity. A case study of mapping the flea index in Mongolian gerbil was conducted to evaluate the effectiveness of this idea and the environmental similarity method. The result showed that compared with multiple linear regression (MLR), the environmental similarity method produced a flea index map of higher accuracy (root mean squared error (RMSE) of 6.6493 and mean absolute error (MAE) 1.2713 than that from MLR (7.293 RMSE and MAE for MLR is not 1.2713 but 2.928). In addition, the spatial distribution of prediction uncertainty estimated from the environmental similarity method can be used as an indicator of spatial distribution of prediction accuracy.