Respira

• 

  Hotspots

• 

  Smoke trajectories

• 

  Chatbot

• 

Inspiration

Wildfires are becoming increasingly rampant and hazardous in the past few decades. Also on the rise are respiratory issues: Asthma prevalence in adults has shown a significant increase in recent years, particularly between 2013 and 2021 (National Institute of Health). When we have wildfires in other states, it can be unexpected to have worsening air quality across the country. Oftentimes, the communication of worsening air quality to the public is last minute. This can affect the elderly and others with respiratory issues, who may not receive this news until they step outside with no protective precautions.

What it does

Areas of high potential fires and effected radius upon actualization of fires Potential smoke/haze directions, carried by wind, rapidly after wildfire starts. In addition, how far the smoke be carried, and obtaining an affected radius from the wildfire epicenter. Chatbot for ease of access to answer demanding questions. Ex. What are the chances of fire and as a result, smoke/haze in my surrounding areas?

How we built it

On the frontend, we used Next.js, React, and Tailwind for styling. The map interface used Google Maps API and various visualization APIs. For our backend, we used FastAPI, Fire Environment Mapping System API, mathematical modeling, and Gemini LLM.

Challenges we ran into

Our challenges were split between the backend and frontend. In the backend, we had trouble with the mathematical modeling. The calculations were not straightforward, and the path with AI assistance eventually allowed us to obtain how far the smoke can be carried, and obtaining an affected radius from the wildfire epicenter. On the frontend, we had trouble with the frontend map visualization. We needed precise information from the backend, and then we needed to create cold/warm/hot zones near the wildfire epicenter.

Accomplishments that we're proud of

We're proud of the mathematical modeling we were able to build on the backend, despite the AI assistance we used and the setbacks this caused us (we could've implemented more personalized health alerts for sensitive groups).

What we learned

We learned how to utilize Gemini LLM to build out a robust backend, processing questions and then outputting responses with contextual information with our data. We also learned about mathematical modeling tools that are able to process complicated data and output significant information for our frontend mapping visualization.

What's next for Respira

Enhanced Prediction Confidence: Achieve even greater reliability and precision in our AI-powered air quality forecasts, ensuring users receive the most accurate data. Personalized Health Alerts: Integrate individual health sensitivities to deliver highly customized warnings and recommendations, especially for vulnerable populations. Travel Route Exposure Prediction: Provide real-time pollutant exposure predictions for planned travel routes, empowering users to choose the healthiest path. Smart Device Integration: Seamlessly sync with smartwatches and other wearables for continuous, on-the-go air quality monitoring and instant alerts.

Built With

• fastapi
• gemini
• google-maps
• next.js
• python
• react
• tailwind