Dengue disease prediction using machine learning algorithms: a review

Dengue fever is a mosquito-borne viral disease spreading in tropical and subtropical regions. It affects approximately 141 countries. Currently, there is no specific medication to treat the disease and supportive care is available. Vaccine is accessible but with many limitations. Because of that, one of the best solutions is to break the dengue cycle. The study aims to design an accurate and timely prediction system that can achieve effective targeting of possible dengue outbreak areas. From the literature review, Support Vector Machine (SVM), Random Forest (RF), Bayes Network (BN) and Bayesian Ridge Regression have shown their outstanding performances in several studies for dengue outbreak forecast system. The most important and commonly used predictors in outbreak forecast are meteorological (rainfall, temperature and relative humidity), entomological (mosquito population size, dengue virus serotypes in mosquitoes) and socioeconomic (region population household income, education status, employment rate, etc.). Both regression and classification models can be employed construct dengue disease spread predictive analysis. In Malaysia, there is a lack of research to compare various machine learning algorithms on dengue outbreak prediction. The goal of this research is to build an optimized model to predict dengue outbreak, so preventive measures can be taken to break the dengue life cycle in Malaysia. The model can help government to track and forecast the outbreak of dengue fever. In future research, other machine learning algorithms and training predictors can be added or modified to obtain a model with better performance.