Year: 2026
Type: preprint
Abstract: Accurate and timely probabilistic dengue forecasts can support public health preparedness, but producing reliable forecasts across locations and time horizons remains challenging. We evaluated five statistical, machine-learning, deep-learning, and foundation-model approaches for forecasting monthly reported dengue cases across 5,570 Brazilian municipalities at 1-, 3-, 6- and 12-month horizons, with forecasts backtested from January 2020 to August 2025. The time-series foundation model TimesFM maintained comparatively stable probabilistic forecast performance at longer horizons, while most other models showed declining performance. We then tested whether spatial representations of human and environmental context from a Population Dynamics Foundation Model (PDFM) could improve forecasts through post-hoc residual adjustment. PDFM adjustment improved mean probabilistic forecast performance, measured by the Weighted Interval Score (WIS), in 19 of 20 model-horizon combinations, although the magnitude and frequency of improvements varied across models, horizons and municipalities. The largest improvements were observed for N-BEATS, including a reduction in mean WIS of 1.13 at the six-month horizon (95% CI [0.84, 1.48]; Diebold-Mariano p < 0.0001), with improvements also observed for XGBoost. The proportion of forecasts improved by PDFM generally increased with locally reported dengue case counts, although this relationship varied between models. Covariate selection, spatial fitting and forecast aggregation strategies showed strong model-dependent effects, while monthly model retraining provided no consistent advantage in the settings evaluated. Together, these findings highlight the potential of pretrained representations for dengue forecasting, with benefits dependent on forecasting architecture, horizon and modelling configuration.
Source: medRxiv
Authors: John-Stuart Brittain, Cathal Mills, Ciara Judge, Leonardo Soares Bastos, Lais Picinini Freitas, Luiz Max Carvalho, Eduardo Correa Araujo, Flavio Codeco Coelho, Yanhua Xu, Benjamin Reddy, Rama Kumar Pasumarthi, Tomer Shekel, Shravya Shetty, Amy Chung-Yu Chou, Anna Carter, Arbaaz Muslim, Gautam Prasad, Kasumi Widner, Monica Bharel, Samuel V Scarpino, Seth Flaxman, Samir Bhatt, Joseph L. H. Tsui, Moritz U.G. Kraemer
Language: English
Field: Medicine
Domain: Health Sciences
Open Access status: green
Awards: G1098891978, G1272572850, G1809983139, ISARIC 3.0: Strengthening the global clinical research ecosystem for epidemic and pandemic prevention and preparedness, Dengue Advanced Readiness Tools (DART) - integrated digital system for dengue outbreak prediction and monitoring, G3106060835, Global.health: data science and sharing for early response to emerging infectious diseases, Mediators of regulation and immunity in helminth infections., G3555387431, G3587002227, G4320171638, ISARIC 3.0: Strengthening the global clinical research ecosystem for epidemic and pandemic prevention and preparedness, University of Oxford - to strengthen clinical research readiness for epidemic disease response including by coordinating and strengthening the conduct of endemic disease ..., Global.health: data science and sharing for early response to emerging infectious diseases, Dengue Advanced Readiness Tools (DART) - integrated digital system for dengue outbreak prediction and monitoring, MOnitoring Outbreak events for Disease surveillance in a data science context, Eco-Epidemiological Intelligence for early Warning and response to mosquito-borne disease risk in Endemic and Emergence settings, G6668181273, Institut National de Recherche Biomédicale (INRB) - To improve public health decision-making in the DRC by strengthening data analytics and modeling capacity for mpox outbreak response, G7701192265, An advanced cookstove intervention to prevent pneumonia in children under 5 years old in Malawi: a cluster randomised controlled trial.
OpenAlex ID: W7225399126