Dengue Risk Forecast with Mosquito Vector: A Multicomponent Fusion Approach Based on Spatiotemporal Analysis.

Area of Science:

• Epidemiology
• Public Health
• Data Science

Background:

• Dengue fever is a significant global health threat requiring accurate forecasting.
• Existing dengue forecasting models often overlook crucial spatial dependencies and temporal patterns.
• Mosquitoes are primary vectors for dengue transmission, necessitating their inclusion in predictive models.

Purpose of the Study:

• To develop an advanced model for dengue risk forecasting that incorporates both spatial and temporal dynamics.
• To address the limitations of current models by considering spatial dependencies and temporal periodicity.
• To improve the accuracy and reliability of dengue fever predictions.

Main Methods:

• Proposed a spatiotemporal component fusion model (STCFM) integrating mosquito abundance and spatiotemporal lags.
• Employed multiscale modeling for temporal dependencies to capture historical data variations.
• Utilized multivariate spatial correlation analysis and ConvLSTM for spatial feature representation and learning.
• Implemented a stacking strategy fusion within ensemble learning for final forecast generation.

Main Results:

• The STCFM demonstrated improved prediction accuracy on real-world dengue datasets.
• The model effectively captured and represented complex spatiotemporal features.
• STCFM outperformed existing candidate models in dengue risk forecasting.

Conclusions:

• The STCFM offers a robust approach to dengue fever forecasting by effectively integrating spatiotemporal factors.
• The model's component construction strategy contributes to its superior performance.
• Accurate dengue forecasting using STCFM can aid public health interventions and resource allocation.