Prediction of the dengue incidence in Bangladesh using machine learning

Although the organisms or insects causing infectious disease are normally harmless, under certain conditions, some may cause serious diseases. Dengue is such kind of disease that is transmitted by mosquitoes and caused by any of the four related dengue viruses and causes infections leading to fever and fatigue. Taking rest and home remedies can sometimes cure mild infections while sometimes life-threatening infections may need hospitalization. As such, the outbreak of dengue is one of the top diseases causing the 105most deaths worldwide including in Bangladesh. Since 1964, Bangladesh has experienced the sporadic occurrence of dengue until 2000 when the first epidemic of dengue was reported in the capital city, Dhaka. Since then, the disease has shown an annual occurrence in all major cities of the country. The state-of-the-art methodologies e.g., machine learning approaches are now being used in many countries for early predicting or forecasting dengue cases. In this paper, we propose to drive machine learning algorithms for the early prediction of dengue cases in Bangladesh. We collect and preprocess meteorological data of Dhaka city to fit into the machine learning models. In this work, we proposed a weighted average ensemble technique of five machine learning methodologies for predicting the number of dengue cases per month from meteorological data. The proposed approach produces promising results in predicting the dengue cases for our testing data samples, which shows a great potentiality to employ machine learning algorithms successfully for early dengue incident prediction in various cities of Bangladesh. We hope that our methodology can contribute to further research on predicting dengue incidents in Bangladesh using meteorological data.