Dengue Outbreak Prediction in Bangladesh Based on Meteorological Factors

초록

Bangladesh witnessed its most severe dengue outbreak in September to October 2023, with an unprecedented number of registered cases, accompanied by 1,272 fatalities. According to the World Health Organization (WHO), unusually heavy rainfall episodes, along with high humidity and temperatures, are potential reasons for dengue outbreaks. An early warning system for dengue outbreaks based on meteorological variables can effectively reduce the severity of these situations. In this study, we create a custom dataset for experimentation, which contains the number of dengue incidents and the meteorological data of Bangladesh from 2008 to 2022, month by month. The proposed model, CXGDBoost, extrapolates imminent dengue cases based on climate data (Rainfall, Humidity, MaxMin temperature). The CXGDBoost model yielded a remarkable R-squared value of 0.946 with an MAE (Mean Absolute Error) of 0.077 and an MSE (Mean Squared Error) of 0.054. The empirical evaluation of the model demonstrated its prominent performance, allowing for more accurate prediction of outbreak occurrences compared with other existing methods. This study presents a rigorous approach for the early forecasting of dengue outbreaks in Bangladesh. The model's high accuracy empowers government officials and health authorities to adopt appropriate preventive measures, potentially mitigating the severity of future outbreaks.

키워드