Hierarchical Bayesian Modeling of Dengue in Recife, Brazil (2015-2024): The Role of Spatial Granularity and Data Quality for Epidemiological Risk Mapping
Fuente:
arXiv
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917016761270272 |
|---|---|
| author | Santos, Marcílio Ferreira dos de Melo, Andreza dos Santos Rodrigues |
| author_facet | Santos, Marcílio Ferreira dos de Melo, Andreza dos Santos Rodrigues |
| contents | Dengue remains one of Brazil's major epidemiological challenges, marked by strong intra-urban inequalities and the influence of climatic and socio-environmental factors. This study analyzed confirmed dengue cases in Recife from 2015 to 2024 using a Bayesian hierarchical spatio-temporal model implemented in R-INLA, combining a BYM2 spatial structure with an RW1 temporal component. Covariates included population density, household size, income, drainage channels, lagged precipitation, and mean temperature. Population density and household size had positive effects on dengue risk, while income and channel presence were protective. Lagged precipitation increased risk, and higher temperatures showed an inverse association, suggesting thermal thresholds for vector activity. The model achieved good fit (DIC=65817; WAIC=64506) and stable convergence, with moderate residual spatial autocorrelation (phi=0.06) and a smooth temporal trend between 2016 and 2019. Spatio-temporal estimates revealed persistent high-risk clusters in northern and western Recife, overlapping with areas of higher density and social vulnerability. Beyond reproducing historical patterns, the Bayesian model supports probabilistic forecasting and early warning systems. Compared with classical models (GLM, SAR, GWR, GTWR), INLA explicitly integrates uncertainty and spatial-temporal dependence, offering credible interval inference for decision-making in urban health management. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_13672 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Hierarchical Bayesian Modeling of Dengue in Recife, Brazil (2015-2024): The Role of Spatial Granularity and Data Quality for Epidemiological Risk Mapping Santos, Marcílio Ferreira dos de Melo, Andreza dos Santos Rodrigues Applications Quantitative Methods Methodology Dengue remains one of Brazil's major epidemiological challenges, marked by strong intra-urban inequalities and the influence of climatic and socio-environmental factors. This study analyzed confirmed dengue cases in Recife from 2015 to 2024 using a Bayesian hierarchical spatio-temporal model implemented in R-INLA, combining a BYM2 spatial structure with an RW1 temporal component. Covariates included population density, household size, income, drainage channels, lagged precipitation, and mean temperature. Population density and household size had positive effects on dengue risk, while income and channel presence were protective. Lagged precipitation increased risk, and higher temperatures showed an inverse association, suggesting thermal thresholds for vector activity. The model achieved good fit (DIC=65817; WAIC=64506) and stable convergence, with moderate residual spatial autocorrelation (phi=0.06) and a smooth temporal trend between 2016 and 2019. Spatio-temporal estimates revealed persistent high-risk clusters in northern and western Recife, overlapping with areas of higher density and social vulnerability. Beyond reproducing historical patterns, the Bayesian model supports probabilistic forecasting and early warning systems. Compared with classical models (GLM, SAR, GWR, GTWR), INLA explicitly integrates uncertainty and spatial-temporal dependence, offering credible interval inference for decision-making in urban health management. |
| title | Hierarchical Bayesian Modeling of Dengue in Recife, Brazil (2015-2024): The Role of Spatial Granularity and Data Quality for Epidemiological Risk Mapping |
| topic | Applications Quantitative Methods Methodology |
| url | https://arxiv.org/abs/2510.13672 |