Prediction of Areas Endemic for Schistosomiasis through Use of Discriminant Analysis of Environmental Data

摘要

Disceminant analysis models were developed to describe the known distribution of schistosomiasis in the Caribbean and in the Philippines in terms of summary variables of monthly mean values of precipitation and temperature. Presence or absence of Schistosoma mansoni was correctly classified with four variables in 87.1 per cent of the known observations in the Caribbean. In the Philippines, presence or absence of Schistosoma japonica was correctly classified in 93.2 per cent of the known observations also with four variables. Using the discriminant equations developed by these two models it was possible to calculate the probability for those locations where disease data were not available, and to classify each of those locations as either disease present or disease absent.