Can open-source AI help us forecast outbreaks? At ARTPARK, Indian Institute of Science, we’re building ForecastHub — an open digital public good that packages the entire disease forecasting workflow into reusable, modular components.

Built initially for dengue in Karnataka, ForecastHub combines climate, vector, and case data to generate district-level 4-week-ahead forecasts — helping health teams deploy fogging, fever clinics, and awareness campaigns in time. But the core idea goes beyond dengue: to create a shared, open infrastructure where developers, researchers, and governments can contribute new models, datasets, and visualizations to improve forecasting across diseases.

In this session, we’ll dive into:
The data factory → model → insight architecture powering our open pipeline
How you can generate your own forecasts using Python packages and config-based pipelines
Publishing results to a shared ForecastHub that supports GitHub contributions, API integrations, and Dockerized deployments
How modular “building blocks” like these can evolve into digital public goods — reusable, interoperable systems for predictive public health
For example, Imagine a malaria researcher plugging in their data to retrain our model, or a civic developer building a custom dashboard for municipal use — all on open infrastructure. ForecastHub is an invitation to the open-source community to code for collective intelligence — to build tools that not only predict risk but enable timely, coordinated action.