FAIR | Devpost

Inspiration

We noticed that respiratory issues are becoming increasingly prominent in our communities - from rising asthma rates in children to COPD diagnoses in adults, and the growing concern about lung cancer linked to air pollution exposure. Our vision was to create a user-friendly app that could help combat theprogression of lung-related diseases while actively promoting healthy respiratory habits.

What it does

Fair is an AI-powered environmental justice platform designed to provide personalized air quality monitoring and health protection for vulnerable populations. The application addresses the growing concern of respiratory health issues by combining individual health profiles with real-time environmental data to deliver customized guidance that helps users protect their lung health and make informed decisions about daily activities.

How we built it

We builtit using React Native with Expo for cross-platform mobile development, providing a foundation that works on both iOS and Android. The frontend uses React Navigation v6 for tab-based navigationand React Context for state management, with specialized contexts for location services, user profiles, and authentication. The backend leverages Supabase as a PostgreSQL-basedbackend-as-a-service, providing real-time subscriptions, email authentication, and a comprehensive database schema with 8 tables fo users, environmental data, community reports, and AI templates. Multiple APIs are integrated including AirNow for EPA air quality data, OpenWeatherMap for weather and pollution concentrations, and ElevenLabs for text-to-speech functionality powering the running coach feature,

Challenges we ran into

Multi-source data synchronization - Combining data from AirNow, OpenWeatherMap, and EPA APIs with different response formats and timing is a technical challenge. React Native context performance bottlenecks - Managing multiple real-time subscriptions through React Context can cause performance issues.

Accomplishments that we're proud of

We are very proud of out robust frontend with great risk assessment model to provide realistic feedback and recommendation along with an interactive coach to help with users respiratory health.

What we learned

We learned how to integrate multiple new APIs like EPA AirNow for real-time air quality data and OpenWeatherMap for meteorological conditions, which was our first time working with government environmental data sources. We also dove into different ML approaches, experimenting with Random Forest models for health risk assessment and Google TimesFM via Hugging Face for time series forecasting, which taught us a lot about ensemble model techniques.

What's next for FAIR

We're planning to expand our dataset for improved ML accuracy and integrate additional pollutant sources like wildfire smoke and indoor air quality monitoring. The next big step is developing partnerships with healthcare providers for pilot testing and conducting user studies to validate that Fair actually improves health outcomes for respiratory patients.