Inspiration
Climate change is reshaping global health. Rising temperatures, shifting rainfall, and changing ecosystems are enabling mosquitoes and ticks to expand into new regions. This increases the risk of diseases like malaria, dengue, and Lyme reaching populations that may be unprepared. We were inspired to build a tool that could not only visualize current risks but also forecast future outbreaks, turning climate data into actionable health insights.
What it does
EpiGlobe is an interactive 3D globe that maps and forecasts vector-borne disease risks worldwide. Users can:
- Explore disease risks (malaria, dengue, Lyme) across different regions and months.
- Filter to see the highest-risk areas globally.
- Forecast future risks using machine learning (XGBoost) to project likely hotspots.
- Visualize it all through a CesiumJS-powered globe.
How we built it
- Data: We integrated climate datasets with disease case reports.
- Modeling: We used XGBoost for time-series forecasting of disease risk.
- Geospatial filtering: We created a land mask to exclude oceans, ensuring realistic predictions.
- Frontend: We built an interactive CesiumJS interface where users can change diseases, months, palettes, and filters in real-time.
- Infrastructure: Served GeoJSON risk layers via a simple local HTTP server for smooth visualization.
Challenges we ran into
- Converting global data into a uniform grid for visualization without overwhelming performance.
- Handling time-series forecasting with limited and noisy disease datasets.
- Getting CesiumJS to properly filter land vs. ocean points.
- Designing a UI that is both powerful and intuitive under hackathon time constraints.
Accomplishments that we're proud of
- Built a fully interactive globe that lets users explore real and forecasted risks.
- Integrated machine learning (XGBoost) to extend predictions into the future.
- Created a flexible filtering UI (color palettes, top-risk filters, dot size) to empower exploration.
- Made the platform generalizable, which makes it easily extendable to other diseases and datasets.
What we learned
- How to preprocess large-scale geospatial datasets into efficient formats.
- The power of combining climate science with machine learning for public health.
- Advanced CesiumJS customization for visualization and UI building.
- The importance of balancing scientific accuracy with usability.
What's next for EpiGlobe
- Expanding to include more diseases (Zika, West Nile, Chikungunya).
- Incorporating real-time satellite data (rainfall, vegetation, temperature).
- Deploying as a web app with cloud hosting and APIs for global health organizations.
- Partnering with NGOs and public health agencies to use EpiGlobe for early-warning systems.
Log in or sign up for Devpost to join the conversation.