Precision Death Forecasting of Dengue Outbreaks in Bangladesh by Blending Statistical and Machine Learning Models

Dengue fever, spread by Aedes mosquitoes, presents a formidable global public health challenge. Despite progress in predictive modeling, the rising incidence and complexity of dengue necessitate more precise forecasting techniques. This study leverages a comprehensive dataset from Bangladesh, spanning 2019 to 2023, to improve dengue outbreak predictions by using hybrid models that blend traditional statistical methods with advanced machine learning algorithms. The research employs a Seasonal Autoregressive Integrated Moving Average (SARIMA) model, along with XGBoost and Random Forest Regression, to predict dengue incidence based on temporal and climatic variables. The dataset consists of 900 daily records, capturing crucial details such as date, month, year, and dengue cases to reflect seasonal trends impacting disease transmission. The results indicate that the XGBoost model, enhanced with features like DayOfWeek and Month, markedly surpasses other models. It achieved the lowest Mean Squared Error (MSE) of 2.6108 and a Root Mean Squared Error (RMSE) of 1.6158, demonstrating superior predictive accuracy. Conversely, the SARIMA model, despite its ability to capture temporal patterns, showed limitations in managing non-linear trends, resulting in a higher RMSE.These findings emphasize the effectiveness of combining machine learning techniques with time series analysis to forecast dengue trends, offering a vital tool for public health officials to apply timely interventions. This study underscores the significance of integrating models and feature engineering to enhance the precision of health predictions, thus contributing to more effective dengue management strategies.