About me

I am an associate professor of Mathematical Epidemiology at the School of Applied Mathematics (EMAp) at FGV, Rio de Janeiro, Brazil. I am also part of the GRAPH network, based at the University of Geneva. In Brazil, I am also one of the coordinators of the Infodengue project. My research interests revolve around the epidemiology of Infectious diseases from the point of view math, statistics and data-science.

I am also a huge enthusiast of open source software and the Python programming language, Further information about this at my GitHub

Research Statement

Arbovirus diseases are among the greatest challenges in global health in recent years. They already claim millions of cases every year, causing enormous suffering, deaths and economic burden. Among these diseases, dengue is perhaps the largest threat, and the fastest expanding arbovirus disease globally.

My research team seeks to help improve the management and control of these diseases, through developing effective and efficient data management protocols and data analysis methodologies based on sound data science principles. Below, we highlight some of our specific research goals:

• Data collection: Our team has developed over the years automated egg-counting software based on computer vision and machine learning, to allow for faster monitoring of Aedes mosquito populations. These software tools are currently being used by the ministry of health throughout the country. This research is conducted in partnership with Fiocruz.
• Public Health data management: Over the past decade, the Infodengue platform, created by us in partnership with Fiocruz, organizes raw surveillance data generated by SUS into a well organized data platform, that generates weekly analytical reports for the public management of dengue, chikungunia and Zika, at federal, state and municipal levels. This platform has been active for over a decade providing value, and serving as an example of the potential of applied research when done in partnership with the public sector.
• Data management and accessibility: Our day to day research depends on quality data. Therefore, we contribute to dissemination and adoption of FAIR principles for data: data should be Findable, Accessible, Interoperable and Reproducible. These principles guide our research and the design of the tools we create. One good example of product developed by our team is the PySUS library, that applies the FAIR principles to the SUS databases, by turning data stored in hard to access government servers, in legacy data formats, into FAIR data, accessible by any data scientist that wants to use this data for their research.
• Epidemiological Modeling: In our group, we are focused on understanding the mechanisms of transmission of infectious diseases. Some of the diseases we have focused our modeling efforts on are mosquito-borne ones, such as dengue, Zika, chikungunya, malaria and yellow fever, but also directly transmitted diseases such as influenza SARS and COVID, and zoonosis such as sporotrichosis and others. Methodologically speaking, we employ a variety of techniques, such as differential equations, agent-based computational models and network models for mechanistic models, but also employ a number of data-driven methods to elicit causal structures, estimate parameters and to make predictions. Some of the data-driven methods we have used include but are not limited to: classical regression analyses, Machine-learning regression and classification models and deep neural networks for predictions.
• Large-scale Epidemiological Analyses: Epidemiological systems are extremely complex systems driven not only by biological/immunological factors, but also by environmental, demographic and climatic ones. Moreover, these systems display a large spatial scope that needs to be studied as a whole, if we hope to understand its systemic properties. In our group, we strive to develop analytical tools that can work at scale, taking in data from all these sources together to generate predictions. One good example of such initiative is our episcanner tool, that continuous monitor the incidence of dengue and chikungunya, to extract transmission parameters from all 5570 Brazilian municipalities.
• Forecasting and Reproducible Science: Our Mosqlimate project, funded by the Wellcome Trust, is focused on producing usable forecasts of dengue burden, that includes all available data sources, but with special attention to the impact of future climate trends. Another innovative aspect of the forecasting platform, is that it keeps track of data, model source code and predictions, in order to guarantee the reproducibility of the entire pipeline.