About

The Infodengue-Mosqlimate Sprint Challenge was a collaborative initiative conducted by the Infodengue and Mosqlimate platform teams, centered around the following guiding question:

What is the expected number of probable dengue cases in the upcoming season, both for Brazil as a whole and by individual Federation Units (UFs)?

The main objective of the Dengue Sprint was to support the development and training of predictive dengue models in Brazil and to provide forecasts for the upcoming season. The specific objectives were to:

• Build a community of modelers aligned in objectives and methodology;

• Train ensemble models using the collective forecasts from all participating teams to generate ensemble predictions;

• Continuously update and monitor model performance.

To facilitate model development, the Mosqlimate platform compiled a comprehensive dataset integrating epidemiological, demographic, and climate data. It also provided a GitHub-based submission template to standardize and aggregate forecasts into an ensemble model.

Six teams participated in the first edition of the Sprint, presenting predictive models during a workshop held on August 16. The outcomes are detailed in a technical report and a scientific article.