Auxiliar files for the manuscript "A statistical model for forecasting probabilistic epidemic bands for dengue cases in Brazil" by Freitas et al.
In 0 getting_data.r we organize the crude data for probable cases from DATASUS and suspected cases downloaded from InfoDengue project using a Mosqlimate API.
The file containing the cases probable and suspected dengue cases by municipality cases.csv.gz with the following variables:
- data_iniSE - Start of the week (Sunday) for the symptoms onset.
- municipio_geocodigo - IBGE 7-digit code for the municipality
- ID_MN_RESI - IBGE 6-digit code for the municipality
- casos - Suspected dengue cases provided by the InfoDengue system, all reported cases.
- casos_prov - Probable cases from SINAN open data, all reported cases excluding descarted cases.
In spatial.tbl.csv there is a correspondence table for health districts (n=118), health regions (n=450) and all IBGE 7-digit code municipalities (n=5570).
In 1_running_models.r we fit INLA model to each health district then predict for the next season. This is done for seasons 2022-2023, 2023-2024 and 2024-2025. We then save the Monte Carlo samples for the predictions for probable and suspected cases for each season, output is available at samples
In 2_replication_analysis.r we provide the code to replicate the tables and figures of the manuscript.