Climate-driven Spatiotemporal Dynamics of Dengue Transmission in Southern Brazil: Integrating Entomological Surveillance and Random Forest Prediction

Abstract Climatic variability is closely linked to dengue transmission; however, its integration with entomological surveillance and machine-learning models remains poorly explored in subtropical regions of South America. This study evaluated the predictive performance of Random Forest models combining meteorological variables, entomological surveillance data, and lagged epidemiological series to predict dengue transmission across four distinct subtropical municipalities in Santa Catarina, southern Brazil, during 2014–2023. Entomo-epidemiological data were obtained from the Diretoria de Vigilância Epidemiológica de Santa Catarina (DIVE/SC), while climate data (temperature and precipitation) were obtained from the Instituto Nacional de Pesquisas Espaciais (INPE) and the National Centers for Environmental Prediction (NCEP). The Artificial Intelligence approach was implemented in Python 3, to predict dengue cases based on climatological elements, vector proliferation (breeding sites), and lagged epidemiological data. Spatiotemporal analysis revealed marked regional heterogeneity: dengue transmission was concentrated in coastal and western zones within optimal vector thermal ranges (22°C–32°C), with epidemic peaks occurring earlier in the western region (weeks 12–16 in Chapecó) than on the coast (weeks 16–20). Therefore, predictive performance varied substantially across regions, achieving high accuracy in Chapecó (R² = 0.82) but lower efficacy in Florianópolis (R² = 0.38). Crucially, the models exhibited a systematic underestimation of extreme epidemic outbreaks when high-incidence historical data were absent from the training sets. These findings demonstrate that regional climatic heterogeneity strongly conditions dengue dynamics, highlighting that locally calibrated predictive models are essential to enhance early warning systems and guide One Health surveillance strategies in subtropical zones.

Authors

Institutions

Publication Details

Journal
Research Square
Published
2026-11-13
DOI
https://doi.org/10.21203/rs.3.rs-11322529/v1
Primary Topic
Mosquito-borne diseases and control
Type
preprint

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

Climate-driven Spatiotemporal Dynamics of Dengue Transmission in Southern Brazil: Integrating Entomological Surveillance and Random Forest Prediction

Mário Francisco Leal de Quadro, Adriano Vitor, Eduardo Augusto Werneck Ribeiro, Michel Nobre Muza et al.
3 links
Research Square
Mosquito-borne diseases and control
preprint

Climate-driven Spatiotemporal Dynamics of Dengue Transmission in Southern Brazil: Integrating Entomological Surveillance and Random Forest Prediction

Mário Francisco Leal de Quadro, Adriano Vitor, Eduardo Augusto Werneck Ribeiro, Michel Nobre Muza, Matheus Ferreira de Souza, Everton Weber Galliani, Caroline Brescian, João Augusto Branche Fuck
preprint en

Abstract

Abstract Climatic variability is closely linked to dengue transmission; however, its integration with entomological surveillance and machine-learning models remains poorly explored in subtropical regions of South America. This study evaluated the predictive performance of Random Forest models combining meteorological variables, entomological surveillance data, and lagged epidemiological series to predict dengue transmission across four distinct subtropical municipalities in Santa Catarina, southern Brazil, during 2014–2023. Entomo-epidemiological data were obtained from the Diretoria de Vigilância Epidemiológica de Santa Catarina (DIVE/SC), while climate data (temperature and precipitation) were obtained from the Instituto Nacional de Pesquisas Espaciais (INPE) and the National Centers for Environmental Prediction (NCEP). The Artificial Intelligence approach was implemented in Python 3, to predict dengue cases based on climatological elements, vector proliferation (breeding sites), and lagged epidemiological data. Spatiotemporal analysis revealed marked regional heterogeneity: dengue transmission was concentrated in coastal and western zones within optimal vector thermal ranges (22°C–32°C), with epidemic peaks occurring earlier in the western region (weeks 12–16 in Chapecó) than on the coast (weeks 16–20). Therefore, predictive performance varied substantially across regions, achieving high accuracy in Chapecó (R² = 0.82) but lower efficacy in Florianópolis (R² = 0.38). Crucially, the models exhibited a systematic underestimation of extreme epidemic outbreaks when high-incidence historical data were absent from the training sets. These findings demonstrate that regional climatic heterogeneity strongly conditions dengue dynamics, highlighting that locally calibrated predictive models are essential to enhance early warning systems and guide One Health surveillance strategies in subtropical zones.

Research Square
Universidade Federal de Santa Catarina (BR), Instituto Federal Catarinense (BR), Instituto Federal de Educação, Ciência e Tecnologia de Santa Catarina (BR)
Fundação de Amparo à Pesquisa e Inovação do Estado de Santa Catarina
Good health and well-being, Climate action
Mosquito-borne diseases and control
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon