A novel approach for dengue outbreak prediction using evolutionary sampling with prediction framework

Area of Science:

• Epidemiology
• Computational Biology
• Data Science

Background:

• Life-threatening viral diseases like dengue fever are increasing globally.
• Dengue outbreak data is often incomplete and inconsistent, hindering accurate prediction.
• Reliable forecasting models are crucial for effective public health interventions.

Purpose of the Study:

• To develop an innovative framework for accurate dengue fever outbreak prediction.
• To address challenges posed by temporal and stochastic dynamics in epidemiological data.
• To improve the reliability of infectious disease forecasting models.

Main Methods:

• The study combined Evolutionary Sampling with Prediction (ESP) and a Minimax K-nearest neighbour imputer.
• A novel Firefly Dynamic Evolution (FDE) approach was used for model parameter optimization.
• A Random Forest classifier was employed to capture complex data relationships, with evaluation via 10-fold cross-validation.

Main Results:

• The proposed model demonstrated low Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) on both local and Brazil dengue datasets.
• Achieved MAE of 22.1 and RMSE of 46.37 on the local dataset.
• Achieved MAE of 48.36 and RMSE of 86.76 on the Brazil dataset, indicating improved accuracy and robustness.

Conclusions:

• The developed model effectively handles data irregularities and missing values in dengue outbreak data.
• The findings suggest significant potential for early warning systems for dengue fever.
• The framework may be applicable to forecasting other infectious diseases.