Forecasting and Early Warning Systems for Dengue Outbreaks: Updated Narrative Review.

Area of Science:

• Public Health
• Epidemiology
• Data Science

Background:

• Dengue fever poses a significant global health threat, necessitating effective prediction and warning systems.
• Existing literature presents diverse methodologies for forecasting dengue outbreaks.

Purpose of the Study:

• To systematically review and synthesize findings on dengue outbreak prediction and warning systems.
• To identify methodologies, key variables, performance, and limitations of current systems.

Main Methods:

• A five-stage review process including literature survey, thematic scope definition, exploratory review, categorization, critical analysis, and narrative synthesis.
• Selection of 14 articles on prediction and 7 on warning systems.
• Analysis of statistical, machine learning, and deep learning models.

Main Results:

• Meteorological and climatic variables are most frequently utilized, followed by epidemiological and entomological data.
• Random Forest and Long Short-Term Memory models show high predictive accuracy for short-term forecasts (up to 1 week).
• Advanced warning systems (e.g., EWARS-TDR, ADSEWS) integrate multiple data sources for longer lead times (up to 13 weeks) compared to classical methods.

Conclusions:

• While advanced models show promise, challenges in data quality, availability, model replicability, and implementation persist.
• Further research and development are needed to enhance the practical application of these systems in public health.
• Integration of diverse data sources and advanced modeling techniques can improve dengue surveillance and response efforts.