A Machine Learning Approach for Dengue Fever Prediction: Case Study of Phayao Province

Abstract

Dengue fever is a serious disease caused by a virus carried by Aedes mosquitoes. It is an important problem of ministries of health in many countries around the world. This research therefore aims to study factors affecting the outbreak of dengue fever and create an effective dengue fever prediction model using machine learning techniques. Data from the Phayao Meteorological Station, including climate, temperature, relative humidity, rainfall, number of rainy days and population data in Phayao Province was collected from the provincial public health database. In addition, the number of dengue fever patients, gender and age group collected from Phayao Hospital between 2017 and 2022 was analyzed and build a model with 10 machine learning techniques. Regression types include Support Vector Machines and Linear Regression. Classification types include Artificial Neural Network, Decision Tree, Naïve Bayes, K-Nearest Neighbors, Deep learning, Random Trees, Gradient Boosting, and Logistic Regression and measure model performance using the 5-Fold Cross Validation method.

All data is created in monthly and weekly datasets. Considering the highest overall accuracy and efficiency. From the performance measurement results of Regression, it was found that Linear Regression provides the best performance with an RMSE of 1.190. From the Classification results, Deep Learning was found to be the most effective model with the highest overall performance, reaching an Accuracy of 99.84%. The results from this research can be used as guidelines for application, especially by various agencies involved in surveillance planning to find areas at risk of spreading and to effectively prevent and control dengue fever.