Outbreak prediction of anthrax in Karnataka using Poisson, negative- binomial and zero truncated models

Abstract

Anthrax is a well–known zoonotic disease caused by Bacillus anthracis, a gram positive, spore forming aerobic bacteria. In livestock Anthrax is usually peracute and highly fatal whereas in humans it is a subacute disease and fatal if not treated immediately. Anthrax is prevalent world–wide and is one of the top ten livestock diseases in India and hence forecasting the outeak of this disease by modelling is very helpful in saving the livestock and thereby averting economic loss to the livestock farmer in particular and country in general. In this paper, the data on the outeak reports of Anthrax in livestock in Karnataka from 2000 to 2014 have been analysed. To develop the model, in the risk factor domain, amount of rainfall, temperature, soil pH, soil type and soil nutrients were collected retrospectively. Further, remote sensed variables like Normalized Difference Vegetative Index (NDVI) and Land surface temperature (LST) were collected using Moderate Resolution Imaging spectro–radiometer (MODIS) tools. Prediction models of anthrax outeaks in Karnataka using count models viz. Poisson, Negative Binomial and Zero truncated models were developed. Initially, prediction models were tested for goodness–of–fit using Chi–square test. Then the models were evaluated using Akaike information criterion (AIC), AIC for correction (AICC) and Baysian Information Criterion (BIC). Zero–Truncated Poisson model provided the best fit for the data on taluk level outeak of Anthrax. An attempt is made to explain how various factors influence outeak of Anthrax. A risk map forecasting the outeaks of Anthrax at taluk level was also generated.