Prediction of Dengue Cases in Bangladesh using Explainable Machine Learning Approach

Dengue fever is transmitted through the wound of an Aedes mosquito that harbors the dengue virus. Transmission occurs when the mosquito acquires the virus from an individual with circulating dengue virus in their bloodstream during a biting interaction. Symptoms typically manifest within four to six days post-infection and can persist for about ten days. While some cases present mild symptoms that could be mistaken for the flu or other viral infections, the potential development of severe complications should not be overlooked. This study aims to develop a machine-learning model capable of utilizing relevant information to predict dengue outbreaks within the geographic regions of Bangladesh. For this purpose, a dataset from Bangladesh's weather forecast has been employed to predict dengue cases across 11 districts. Seven machine learning, including three ensemble learning techniques, have been applied in this work to forecast dengue cases and among these, the supper vector regression model with optimized hyperparameters achieved the best results, illustrating the lowest mean absolute error of 4.112. The prediction results obtained from the machine learning models have been analyzed using a Shapash-based explainable AI framework. Overall, the proposed data-driven analysis underscores the significant potential of machine learning algorithms in predicting dengue epidemics.