Predicting Areas Endemic for Schistosomiasis Using Weather Variables and a Landsat Data Base

Abstract

A system to predict the occurrence of schistosomiasis in the Philippines based on the integration of weather variables and Landsat data is described. An interpolation algorithm was developed to provide mean monthly temperature and precipitation data for any given site. Using the discriminant analysis model, the probability of disease occurrence can then be calculated for any given site. Means and variances from Landsat data were used in a regression model, to inject geographic variables into the system and so improve its predictive capabilities. A disease distribution map was produced, based on the statistical correlation between the probabilities of occurrence and the geographic variables.