PNNL

Abstract

Background Dengue fever is the fastest-growing mosquito-borne viral disease globally, with unprecedented epidemics reported across the Americas in 2024. Brazil and Panama experienced record transmission that exceeded historical trends and strained public health systems, highlighting limitations in existing surveillance and early warning approaches. Although dengue transmission is strongly influenced by climatic and environmental variability, operational forecasting systems capable of integrating heterogeneous data across spatial and temporal scales remain limited. Methods We developed an automated forecasting framework integrating epidemiological surveillance, climate, environmental, and demographic data to predict dengue incidence and spatial risk across Brazil and Panama. Monthly case data from 2015–2024 were combined with multi-source environmental predictors and harmonized through a reproducible data pipeline. Multiple statistical and machine learning models were evaluated within a comparative ensemble framework to generate forecasts and spatial risk maps. Predictive performance was assessed using standard error metrics, and model outputs were compared with observed transmission patterns during the 2024 epidemic period. Findings The framework identified substantial spatial heterogeneity in dengue transmission and successfully forecasted regions experiencing elevated risk during the 2024 epidemic. Ensemble approaches improved predictive performance relative to individual models, with temperature, precipitation variability, and urban environmental indicators emerging as dominant drivers of transmission risk. Forecast outputs demonstrated substantial agreement with observed risk classifications and geographic transmission patterns. Ensemble models reduced forecasting error by approximately 20% in Brazil and 36% in Panama relative to the best-performing alternative models, with larger improvements observed during peak transmission periods. Interpretation These findings demonstrate that automated, scalable forecasting systems integrating environmental and epidemiological data can support proactive identification of high-risk regions and improve dengue preparedness. Implementation of such frameworks within routine surveillance systems may enable earlier intervention, more efficient resource allocation, and improved response to climate-sensitive vector-borne disease outbreaks across the Americas.