Exploration of machine learning models to predict the environmental and remote sensing risk factors of haemonchosis in sheep flocks of Rajasthan, India.

Abstract

Globally haemonchosis in sheep is a known devastating disease imposing considerable economic loss. Understanding the environmental risk factors and their role is essentially required to manage the disease successfully. In this study, 14 years' disease data was analysed to predict the risk factors responsible for the occurrence of the disease. Season-wise analysis revealed high incidence during monsoon and post-monsoon and least in winter and summer seasons. The linear discriminant analysis (LDA) revealed the significant environmental and remote sensing risk factors contributing to haemonchosis incidence as enhanced vegetation index, leaf area index, potential evapotranspiration and specific humidity. Further, significant ecological and environmental risk factors identified using LDA were subjected to the climate-disease modelling and risk maps were generated. Basic reproduction number (R) was estimated and was ranged from 0.76 to 2.08 for >1000 egg per gram of faeces (EPG) in four districts whereas Rvalues of 1.09-1.69 for >2000 EPG in three districts indicating the severity of the infection. The random forest and adaptive boosting models emerged out as best fitted models for both the EPG groups. The results of the study will help to focus on high-risk areas of haemonchosis in sheep to implement the available control strategies and better animal production globally.