InfoDengue is an early-warning system for arbovirosis in Brazil. It relies on the continuous analysis of online, climate and epidemiological data.

In 2021, the system reached the national level with the support of the Brazilian Health Ministry. As a result, all municipalities began to receive weekly InfoDengue bulletins. Reports at the state level can be requested by the state health secretariats. Access the tab “how to participate” to know more. Implemented in 2015, the system was developed by researchers from the Scientific Computing Program (Fundação Oswaldo Cruz, RJ) and the School of Applied Mathematics (Fundação Getúlio Vargas) with the strong collaboration of the Rio de Janeiro Municipal Health Secretariat, the Observatory of Dengue (UFMG) and researchers from the Federal University of Paraná and the State University of Western Paraná.

The Infodengue website is accessed by many people and it is common for us to receive news that this information is used in the definition of travel and other activities.

Our network of collaborators has grown and now includes international research groups seeking solutions for arboviruses in other countries and civil society organizations. By sharing experiences and methodologies, we expand our capacity to innovate on arbovirus surveillance.

The several technical and scientific challenges faced during the development of InfoDengue resulted in theses and dissertations. As the challenges do not end, we continue and will always continue to investigate new solutions to generate useful information for the management and advancement of knowledge of arboviruses and their control.

The InfoDengue system is a semi-automated data collection, harmonization and analysis pipeline that generates indicators of the epidemiological situation of dengue and other arboviruses at the municipal level.

Where does the data come from?

• Reported cases of dengue, zika or chikungunya. These are notifiable diseases, that is, the health worker diagnosing a suspicious case needs to fill out a notification form that feeds a municipal database which is then consolidated at the state level and finally, federally by the Ministry of Health. Only a fraction of these cases are laboratory confirmed, most receive final classification based on clinical and epidemiological criteria. From the notified cases, the incidence indicators that feed the InfoDengue are calculated.

• Weather data: The transmission of arboviruses is greatly influenced by the climate. The transmitting mosquito, Aedes aegypti, requires high temperature and humidity to reproduce and live. The virus infecting the mosquito will also reproduce better at higher temperatures. Temperature and humidity data are obtained from airport weather stations as well as satellite images.

• Demographic data: Epidemiological indicators require population size. Demographic data of Brazilian municipalities are updated each year in Infodengue, using estimates IBGE.

What to do?

• Nowcasting: Disease notification data are always delayed for a variety of reasons including the time between getting sick and seeking medical attention, the time it takes for the healthcare professional to report, and the time it takes to aggregate information into databases. This delay makes decision making difficult because one always has to look at outdated data. To circumvent this problem, a Bayesian inference-based statistical method was developed that combines the delay estimate with the historical temporal pattern of incidence to calculate the expected number of cases each week. This process of nowcasting is described Bastos, Economou et al (2019) ”A modelling approach for correcting reporting delays in disease surveillance data”. A modelling approach for correcting reporting delays in disease surveillance data”. The figure below (taken from Bastos et al 2019) shows in black the true incidence, in red, that observed late and in gray the nowcasting. For further technical details, see the publication.



• Sustained Transmission Detection: Estimation of reproductive number is important in many ways. Sequence of weeks with critical transmission (Rt> 1) indicates sustained transmission. That is, it indicates the phase of spread of the disease, and that appropriate control actions need to be implemented. On the other hand, sub-critical periods (Rt <1) indicate that there is no sustained transmission. To estimate Rt from the incidence data, we developed our own methodology, described in Codeço et al. 2018 "Estimating the effective reproduction number of dengue considering temperature-dependent generation intervals".

• Receptivity Analysis: The transmission of arboviruses requires appropriate weather conditions, which may vary from place to place. In Rio de Janeiro, for example, a temperature higher than 22 degrees is a necessary condition for sustained transmission. In Fortaleza, maximum humidity above 85% is a necessary condition. We have adjusted decision tree models to evaluate which combination of climate and environmental variables are indicative of transmission conditions. This information serves as an early warning of the arrival of the transmission season.

• Atypical Incidence Detection: To estimate epidemic thresholds for dengue, we adapted the Moving Epidemics Method (MEM) Vega et al. 2012, which allows estimating of various quantities of interest based on local historical data such as pre-threshold. - epidemic, typical activity levels, expected values ​​for each week, etc. Because this threshold is derived from the data history, locations with different case volumes have different thresholds. The methodology also provides estimates for the expected number of cases per week, together with confidence intervals that allow the definition of weekly activity zones. These regions are defined similarly to activity levels, but are calculated based on the typical case distribution for each week. This construction allows us to visually identify seasonal patterns, since they present the typical temporal evolution based on the median of expected cases for each week and lower and upper confidence intervals (we use 10% and 90%, respectively), defining activity corridors as a control diagram.

Disease code

Implementation
All code generated for InfoDengue, mostly R and Python, is available at:
​https://github.com/AlertaDengue

2018-2020
EVIGILANCIA: Continuous & integrated arboviruses monitoring

Evigilancia aimed to consolidate the alert system for arboviruses based on hybrid data generated through integrated data analysis, mined from web climatic and epidemiological data, the InfoDengue.
Financiamento: INOVA Fiocruz

2020-2021
Predicting epidemics of mosquito-borne diseases

This project aims to provide medium-term projections for the incidence of new cases of mosquito-borne diseases in Brazil. The project adapts to the Brazilian reality, the experience implemented in Vietnam, where we propose the use of Bayesian space-time hierarchical models to predict the incidence of dengue cases, and the support of auxiliary variables, such as climatic variables (Colon-Gonzalez, Bastos, et al. al. 2021). In this project, we will extend the predictive modeling of dengue in Vietnam to mosquito-borne diseases, in which we will separately explore arboviruses in Brazil and malaria in the Amazon region, which are diseases of great impact and relevance in the country's public health. The models to be proposed will be built and validated using open data, where medium-term projections will be made, that is, six to twelve months ahead for the health microregions. The project will prioritize the use of open data, in addition to the use and development of codes in free environments, so that epidemiological surveillance teams can replicate and adapt according to their needs. In addition to the projections, a specific dashboard for each disease will be developed in parallel, so that the forecasts are accessible to decision makers and the general population.
Financiamento: INOVA Fiocruz

2022-2023
ARBOFRONTEIRA: Predicting epidemics of mosquito-borne diseases in the southern arc of the country

The objective of this proposal is to generate indicators for the surveillance of arboviruses that are specific to the southern border of the country and guidelines for its use in strengthening the surveillance of these municipalities. For this, it is necessary to understand the process of expansion of arboviruses in the three countries and to identify future expansion scenarios. Per Through statistical and mathematical models, we will investigate possible scenarios for the spread of arboviruses and the importance of binational and trinational protocols to the highest precision of control actions. From a practical point of view, models will be applied of nowcasting and prediction at two points: the triple border Foz do Iguaçu, Iguazu and city of the east; and on the border formed by Barracão (PR), D Cerqueira (SC) and Bernardo de Irigoyen (Arg). The models will be incorporated into Infodengue and will be made available specific reports for each region.
Financiamento: INOVA Fiocruz

C. Codeco, F. Coelho, O. Cruz, S. Oliveira, T. Castro, L. Bastos, Infodengue:
A nowcasting system for the surveillance of arboviruses in Brazil, Revue d'Épidémiologie et de Santé Publique, Vol 66, Suppl 5, 2018, Page S386,​https://doi.org/10.1016/j.respe.2018.05.408.

@article{CODECO2018S386,
title = "Infodengue: A nowcasting system for the surveillance of arboviruses in Brazil",
journal = "Revue d'Épidémiologie et de Santé Publique",
volume = "66",
pages = "S386",
year = "2018",
note = "European Congress of Epidemiology “Crises, epidemiological transitions and the role of epidemiologists”",
issn = "0398-7620",
doi = "https://doi.org/10.1016/j.respe.2018.05.408",
url = "http://www.sciencedirect.com/science/article/pii/S0398762018311088",
author = "C. Codeco and F. Coelho and O. Cruz and S. Oliveira and T. Castro and L. Bastos"}

2016

• Coelho et al. Epidemiological data accessibility in Brazil, The Lancet ID, link

2017

• Lana R.M., Gomes M.F.C., Lima T.F.M., Honório N.A., and Codeço C.T. The introduction of dengue follows transportation infrastructure changes in the state of acre, brazil: A network-based analysis. PLOS Neglected Tropical Diseases. link

2018

• Santos, B.C., Coelho F.C., Armstrong M, Sarraceni V, Lemos C. Zika: an ongoing threat to women and infant. Cadernos de Saude Publica. link
• Codeço, C.T., Villela A.M.D. , Coelho F.C. Estimating the effective reproduction number of dengue considering temperature-dependent generation intervals. Epidemics. link
• Codeço, C.T., Coelho F.C., Cruz, O.G., Oliveira, S. Castro, T. Bastos, L.S. Infodengue: A nowcasting system for the surveillance of arboviruses in Brazil. Revue d'Épidémiologie et de Santé Publique. link
• Lana R.M., Morais M.M., Lima T.F.M., Carneiro T.G.S., Stolerman L.M., Santos J.P.C., Cortês J.J.C., Eiras A.E., and Codeço C.T. Assessment of a trap based aedes aegypti surveillance program using mathematical modeling. PLOS ONE, link

2019

• Bastos LS, Economou T, Gomes MG, Villela DAM, Coelho OG, Stoner O, Bailey T, Codeço CT. A modelling approach for correcting reporting delays in disease surveillance data. Statistics in Medicine, link
• Coelho FC, Codeço CT. Precision epidemiology of arboviral diseases. Journal of Public Health and Emergency, link

2020

• Elisa Mussumeci, Flávio Codeço Coelho. Large-scale multivariate forecasting models for Dengue - LSTM versus random forest regression, Spatial and Spatio-temporal Epidemiology. link
• Lowe R., Lee S., Lana R.M., Codeço C.T., Castro M.C., and Pascual M. Emerging arboviruses in the urbanized amazon rainforest. BMJ, link

2021

• Alves L.D., Lana R.M., and Coelho F.C. A framework for weather-driven dengue virus transmission dynamics in different Brazilian regions. International Journal of Environmental Research and Public Health, link
• Colon-Gonzalez, Bastos, et al. (2021) Probabilistic seasonal dengue forecasting in Vietnam: A modelling study using superensembles PLOS Medicine. link

2022

• Codeco C.T., Oliveira S.S., Ferreira D.A.C., Riback T.I.S., Bastos L.S., Lana R.M., Almeida I.F., Godinho V.B., Cruz O.G., and Coelho F.C. Fast expansion of dengue in brazil. The Lancet Regional Health - Americas, link
• Miller S, Preis T, Mizzi G, Bastos LS, Gomes, MFC, Coelho FC, Codeço CT, Moat SH. Faster indicators of chikungunya incidence using Google searches Plos Computational Biology, link

• E-vigilância 2021: ​ http://e-vigilancia.dengue.mat.br/

• If your city in Brazil is not Infodengue, contact the Health Department.
• If you are a student or researcher and would like to contribute analysis and models, write us.
• If you would like to contribute sponsorships, write us.
• If you want to learn how to use infodengue, have suggestions, criticism or if you noticed any errors on the site, write us.

Overcoming arboviruses is everyone's job, all collaboration is welcome!