NafasLokal - Air Exposure Passport 🌬️

"Breathe Smarter. Live Healthier. One Commute at a Time."

Team Member Name

Role	Name
💻 Lead Developer	Aiman
🎤 Pitch Architect	Zaim
🔍 Devil's Advocate	Alif
💡 Ideation Lead	Isyraf
🧪 QA Engineer	Shah

Project Info

NafasLokal is an AI-driven air quality monitoring platform designed specifically for Malaysian urban commuters. We transform scattered pollution data into actionable health decisions through real-time monitoring, AI predictions, and gamified behavior change.

🛠️ Tech Stack

Frontend: Next.js 16 (App Router), React, Tailwind CSS
Backend: Convex (self-hosted real-time database)
AI/ML: Custom predictive models for 24-hour air quality forecasting
Maps: Leaflet with custom heatmap visualization layers
Data Sources: DOE Malaysia, WAQI API, OpenAQ (with web scraping fallback)

🎯 SDG Alignment

- SDG 3: Good Health & Well-Being

Project Idea

💡 Inspiration

Every day, millions of Malaysians commute through polluted urban corridors in the Klang Valley. The air they breathe contains dangerous levels of PM2.5, NO₂, and CO – invisible killers linked to respiratory disease, cardiovascular problems, and reduced life expectancy.

The shocking truth? The data already exists. Government agencies and international organizations collect air quality readings every hour. But this data sits in scattered databases, presented in technical formats that everyday citizens can't use.

We asked ourselves: What if we could turn this data into life-saving decisions?

That's how NafasLokal was born – your personal Air Exposure Passport that protects you and your family, one commute at a time.

🔥 Challenges We Faced

The API Struggle – A First for Us

This hackathon threw us a curveball we've never experienced before: no provided APIs.

In every previous hackathon we've participated in, organizers provided ready-to-use APIs or datasets. This time? We were on our own.

Finding reliable air quality data for Malaysia was incredibly difficult:

DOE Malaysia doesn't offer a public API – we had to web scrape their official portal
WAQI API has rate limits and incomplete Malaysian coverage
OpenAQ data for Malaysia is sparse and often delayed

We spent the first several hours just figuring out how to get data:

// Our web scraping solution for DOE Malaysia
const scrapeDOEData = async (stationId) => {
  // Parse HTML from DOE portal
  // Extract PM2.5, NO2, CO readings
  // Handle inconsistent data formats
  // Implement caching to avoid rate limits
};

Lesson learned: Real-world data is messy, incomplete, and hard to access. But that's exactly why solutions like NafasLokal are needed!

Other Challenges

Multi-source data fusion: Merging data from 3+ sources with different formats, units, and update frequencies
Real-time sync: Ensuring the dashboard updates live without overwhelming the APIs
Health scoring algorithm: Creating a scientifically-grounded risk score from raw pollutant values
Map performance: Rendering heatmaps with thousands of data points smoothly

🏗️ How We Built It

Phase 1: Data Pipeline

We built a robust data aggregation layer that:

Pulls from DOE Malaysia (web scraping)
Queries WAQI API (with fallbacks)
Fetches OpenAQ data (supplementary)
Merges and normalizes into a unified format
Caches results to reduce API calls

Phase 2: Core Platform

Real-time Dashboard: GPS-based location detection → instant air quality display
Interactive Heatmap: Leaflet maps with color-coded pollution density
Health Scoring: Custom algorithm weighing PM2.5, NO₂, and CO levels

The health score formula considers WHO guidelines:

\( \text{Score} = 100 - \left( \frac{PM_{2.5}}{2} + \frac{NO_2}{2.5} + CO \times 8 \right) \)

Phase 3: AI Health Engine

24-hour pollutant forecasts using historical pattern analysis
Personalized health scores based on user profiles (age, conditions)
Vulnerable group advisories for children, elderly, asthma patients

Phase 4: Gamification (BreathQuest)

Daily/weekly challenges to encourage healthy choices
XP system, streaks, and badges
Leaderboards to foster community engagement

📚 What We Learned

Data accessibility is a real problem – Even public health data isn't always publicly accessible
Web scraping is a necessary skill – When APIs don't exist, you build your own
User experience matters – Raw data means nothing if people can't act on it
Gamification drives behavior – Points and badges actually motivate healthy choices
Malaysia needs local solutions – Global apps don't understand haze season or local traffic patterns

🚀 Key Features

Feature	Description
Live Dashboard	Real-time AQI with multi-source data fusion
AI Health Score	Personalized assessment: \( 0 \leq S \leq 100 \)
Pollutant Forecast	24-hour predictions for PM2.5, NO₂, O₃, CO
AI Chatbot	Natural language health assistant
BreathQuest	Gamification with challenges & leaderboards
Heatmap Visualization	Color-coded pollution density maps
Vulnerable Alerts	Targeted advisories for sensitive groups

🌏 Impact & Vision

Immediate Impact:

Reduce individual pollution exposure by up to 30% through smarter route choices
Protect vulnerable populations with timely, personalized alerts
Create behavioral change through gamification

Future Vision: