Machine Learning-Based Approaches for Location Based Dengue Prediction: Review

Abstract

Dengue is a fast-spreading viral disease which has no preventive medicine. Due to this infectious disease, almost half of the global population is at risk. Consequently, much research has been conducted using various medical as well as computational methods in order to prevent this menace. The main aim of this paper is to review machine learning approaches to this problem and to identify the most suitable method to predict the spread of this disease for distinctive geographical areas of countries like Sri Lanka. We consider environmental factors such as climate and vegetation data, dengue case data along with the population of a specific geographic area for the disease outbreak predictions. Specifically, this paper consists of the following sections: (i) A brief description of the disease and the factors affecting the spread; (ii) review the pattern of the environmental and population factors affecting the spread; (iii) a review and comparison of machine learning algorithms for prediction of the spread of the disease (SVM, decision tree, neural network, and random forest).