BreatheSafe

Breathe Better with BreatheSafe

Comment

Inspiration

Air pollution is a growing global health crisis, yet awareness about its immediate impact remains low. We were inspired to build a solution that not only tracks pollution but empowers individuals and communities with real-time data, personalized health insights, and a direct line of engagement with authorities. Our goal was to go beyond AQI dashboards and create an intelligent, citizen-centric environmental assistant.

What it does

BreatheSafe is an AI-powered air quality tracker that:

  • Displays a real-time air quality dashboard (AQI, PM2.5, PM10, CO, SO₂, etc.)
  • Predicts health risk scores using machine learning
  • Suggests pollution-aware travel routes using Google Maps API + AQI overlays
  • Sends real-time health warnings in high AQI zones
  • Generates daily natural language summaries via LLMs (e.g., Gemini)
  • Offers an LLM-powered chatbot for pollution and health queries
  • Detects pollution anomalies using unsupervised ML
  • Enables community pollution reporting directly to CPCB
  • Hosts an interactive "Air Quality Academy" with quizzes and certificates
  • Gives personalized, data-driven health recommendations

How we built it

We used the following technologies:

  • Frontend: React.js + TypeScript, styled with Tailwind CSS and Framer Motion for a clean, animated interface.
  • Backend: Node.js + Express.js for server-side logic and APIs.
  • Database: MongoDB to store user profiles, AQI logs, and reports.
  • Authentication: Google OAuth 2.0 and JWT for secure login/session handling.
  • AI/ML:
    • Random Forest for health risk prediction
    • Isolation Forest for anomaly detection in air quality
  • LLMs:
    • Botpress for chatbot queries
    • Google Gemini for AI-generated summaries
  • APIs: WAQI API + GPS for real-time air data, Google Maps API for routing, and CPCB webhook/API for issue reporting.

Challenges we ran into

  • Ensuring real-time synchronization between user location, AQI data, and health warnings
  • Balancing LLM integration with performance and user data privacy
  • Designing a meaningful user experience that is both informative and engaging
  • Handling pollution anomaly detection with limited labeled data
  • Setting up authenticated reporting to government portals via APIs

Accomplishments that we're proud of

  • Successfully integrated AI and LLMs for meaningful, real-world pollution awareness
  • Enabled real-time, pollution-aware route planning — a unique feature among similar apps
  • Built a working CPCB reporting feature to close the citizen-to-government feedback loop
  • Developed an Air Quality Academy to educate and reward users for learning
  • Created an end-to-end pipeline from data ingestion to personalized insight delivery

What we learned

  • How to combine environmental data with user behavior and ML to drive actionable insights
  • The importance of clean UI/UX in delivering serious health-related content
  • How to securely integrate third-party APIs like Google Maps and WAQI
  • Best practices in deploying scalable Node.js apps with real-time capabilities
  • The power of LLMs in summarizing data and engaging users

What's next for BreatheSafe

  • Deploying a mobile app version to enhance accessibility
  • Adding community features like pollution heatmaps and group alerts
  • Expanding reporting integration with more government and civic bodies
  • Fine-tuning ML models with more granular health data and feedback
  • Partnering with schools and local organizations for awareness campaigns

Built With

Share this project:

Updates