Predicting dengue cases through Machine Learning and Deep Learning: a systematic review

• Autor(a) principal: Batista, Ewerthon Dyego de Araújo
• Data de Publicação: 2021
• Outros Autores: Araújo, Wellington Candeia de, Lira, Romeryto Vieira, Batista, Laryssa Izabel de Araujo
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Research, Society and Development
• Texto Completo: https://rsdjournal.org/index.php/rsd/article/view/19347
• topic: Dengue; Forecast; Machine Learning; Deep learning.; Dengue; Pronóstico; Apndizaje automático; Aprendizaje profundo.; Dengue; Previsão; Machine Learning; Deep learning.

Introduction: dengue is an arbovirus caused by the DENV virus and transmitted to humans through the Aedes aegypti mosquito. Currently, there is no effective vaccine to combat all serology of the virus. Therefore, the fight against the disease turns to preventive measures against the proliferation of the mosquito. Researchers are using Machine Learning (ML) and Deep Learning (DL) as tools to predict cases of dengue and help governments in this fight. Objective: to identify which ML and DL techniques and approaches are being used to predict dengue. Methods: systematic review carried out on the bases of the areas of Medicine and Computing in order to answer the research questions: it is possible to make predictions of dengue cases using ML and DL techniques, which techniques are used, where the studies are being performed, how and what data is being used? Results: after performing the searches, applying the inclusion, exclusion and in-depth reading criteria, 14 articles were approved. The Random Forest (RF), Support Vector Regression (SVR), and Long Short-Term Memory (LSTM) techniques are present in 85% of the works. Regarding the data, most were used 10 years of historical data on the disease and climate information. Finally, the Root Mean Absolute Error (RMSE) technique was preferred to measure the error. Conclusion: the review showed the feasibility of using ML and DL techniques to predict dengue cases, with a low error rate and validated through statistical techniques.