Abstract:
Malaria is the most prevalent mosquito-borne disease throughout tropical and subtropical regions of the world with severe medical, economic, and social impact. Malaria is a serious public health problem in Ethiopia since 1959, even if, its morbidity and mortality have been reduced starting from 2001. Various studies were conducted to predict the malaria epidemic using mathematical and statistical approaches, nevertheless, they had no learning capabilities. In this paper, we present a Type-2 Fuzzy Logic Based System for Malaria epidemic prediction in Ethiopia which was trained using real data collected throughout Ethiopia from 2013 to 2017. Fuzzy Logic Based Systems provide a transparent model which employs IF-Then rules for the prediction that could be easily analyzed and interpreted by decision-makers. This is quite important to fight the sources of Malaria and take the needed preventive measures where the generated rules from our system were able to explain the situations and intensity of input factors which contributed to Malaria epidemic incidence up to three months ahead. The presented Type-2 Fuzzy Logic System (T2FLS) learns its rules and fuzzy set parameters from data and was able to outperform its counterparts T1FLS in 2% and ANFIS in 0.33% in the accuracy of prediction of Malaria epidemic in Ethiopia. In addition, the proposed system did shed light on the main causes behind such outbreaks in Ethiopia because of its high level of interpretability
