A systematic review of dengue classification models using machine learning

Maquen-Niño G.L.E.; Bravo J.; Alarcón R.; Adrianzén-Olano I.; Vega-Huerta H. · RISTI - Revista Iberica de Sistemas e Tecnologias de Informacao (2023) · DOI: 10.17013/risti.50.5-27

Resumen

Dengue is an arboviral disease that annually reports a large number of infected on the north coast and the Peruvian jungle. According to statistics, it is increasing yearly. This article aims to develop a systematic review of the scientific literature on the study variables and the machine learning methods currently used for detecting dengue infection. The methodology used was PRISMA, initially mapping the literature of 274 scientific articles, leaving 33 articles selected for the systematic review. The results obtained are that the most used machine learning algorithms are neural networks (NN) and support vector machine (SVM). Likewise, it has been found that scientists tend to carry out research with climatic or demographic variables to obtain better results. It is concluded that the machine learning methods that have been used the most are neural networks of different types: convolutional, recurrent, deep, and multilayer, and for the prediction of dengue outbreaks the time series methods with LSTM and ARIMA were the predominant ones, it was also established that the trend is towards the inclusion of climatic and demographic variables in the prediction models.

Instituciones: Universidad Nacional Mayor de San Marcos; Universidad Nacional Pedro Ruiz Gallo; Universidad Nacional Toribio Rodríguez de Mendoza de Amazonas