Every breath we take tells a story about our planet.
For millions of people around the world, that story has become one of silent danger — the air they breathe is slowly harming their health.
According to the World Health Organization (WHO), nearly 99% of the global population is exposed to air that exceeds safe pollution levels.
Yet, despite the urgency, real-time and accessible air quality information is still lacking for most communities.
That’s where BreathSafe comes in. 🌱
BreathSafe, developed by Team TEMPO 11, transforms NASA’s atmospheric science into an intelligent, user-centered health protection system.
By integrating real-time data from the NASA TEMPO (Tropospheric Emissions: Monitoring of Pollution) mission with ground-based sensors, public health datasets, and AI analytics, we help people understand, anticipate, and act upon air quality conditions that affect their lives.
A dynamic interface combining real-time air quality data with personalized insights.
Visualizes pollutants such as PM2.5, NO₂, and O₃ through an intuitive map, combining NASA satellite and local ground sensor data.
Uses real-time geolocation to display precise air quality information for each user’s exact position.
Sends instant notifications when pollution levels change — allowing users to plan outdoor activities safely.
Delivers personalized health guidance for groups such as children, elderly, pregnant users, and those with respiratory conditions.
Provides daily eco-health suggestions and protective habits based on real-time air quality.
- NASA TEMPO: high-resolution air pollution data (O₃, NO₂, CO, CH₂O, PM, Aerosol Index)
- Meteomatics API: real-time weather parameters (temperature, humidity, wind speed, radiation, precipitation)
- IPCC CIESIN SRES Dataset: historical and projected emissions data
- Spatial harmonization with WGS84 / EPSG:4326
- Bilinear interpolation & nearest neighbor sampling
- Data normalization with quality flag validation
- Missing data recovery using exponential decay and median neighborhood fill
- Mamdani-type Fuzzy Logic System with 100+ expert rules
- Fuzzy sets: Low → Moderate → High → Extreme
- Triangular & Gaussian membership functions
- Centroid defuzzification producing final air-quality class (Good → Hazardous)
- Qwen2.5-1.5B Transformer (Small Language Model)
- Generates bilingual (EN/PT) explanations and recommendations
- Low-latency reasoning (<800 ms) for real-time insights
- Automates environmental reports and summary generation
- Progressive Web App (PWA) collecting geolocation data
- Serverless backend with autoscaling & sub-second response
- REST API endpoints for inference and explainability
- Secure infrastructure using key vaults, containers & OpenTelemetry
- Cross-validation using AERONET and urban monitoring stations
- Metrics: F1 (unhealthy+), macro accuracy, weighted kappa, rank correlation
- Bayesian optimization to fine-tune fuzzy thresholds
✅ Integration of NASA satellite, ground, and weather data
✅ Near real-time monitoring and forecasting
✅ Validated accuracy across multiple data sources
✅ Historical trends and seasonal visualizations
✅ Health-based alerts and proactive recommendations
✅ User-tailored insights by health and location profile
🚀 Digital Phenotypes — analyze human and community responses to air conditions
🤖 Edge AI — local, privacy-preserving, real-time analysis
🎯 Hyper-Personalization — adapt to exposure, habits, and local environments
🧩 Causal AI Models — move from correlation to cause-and-effect relationships
💬 Explainable AI (XAI) — transparency in why alerts and predictions are made
🌐 System Integrations — future partnerships with other mobile and IoT systems
We applied AI throughout the project lifecycle:
| Application | Tools |
|---|---|
| Code generation & system logic | ChatGPT, Manus, Claude, Gemini |
| Text & data interpretation | Qwen2.5 |
| Image & video generation | Manus, Gemini, Google Labs Flow (VEO 3 Quality) |
| Personalized communications | OLLAMA / Qwen2.5 |
| Source | Type | Role |
|---|---|---|
| NASA TEMPO / EARTHDATA / AIR QUALITY | Satellite | Atmospheric pollutants |
| Meteomatics | API | Weather data |
| IPCC CIESIN SRES | Dataset | Emission projections |
| WHO / EPA / AirNow / NEJM / EHP / Springer | Validation | Health and environmental research |
| GitHub | Platform | Repository & version control |
- Project: BreathSafe
- Theme: Data for Health & Environment
- Mission: Empower communities to understand and act upon air quality conditions through real-time NASA data and AI-driven insights.
Team TEMPO 11 — BreathSafe Project
🌐 www.tempo11.earth
✉️ [email protected]
“We don’t just monitor the air. We make it understandable, actionable, and personal.” 🌬️