RespiGuard: AI-Powered Respiratory Health Assistant

created & enhanced during Hack Zenith 2026

“Others measure the air; we measure its impact on you.”

A personalized medical digital twin that combines real-time hyper-local air quality data with AI-driven clinical guidance to protect respiratory patients.

---

🚩 The Problem

• Generic Data: Standard weather apps show city-wide averages, ignoring the pollution right outside your door.
• Panic, Not Pragmatism: Global AQI standards often flag Indian air as "Hazardous" 24/7, causing alarm fatigue without actionable advice.
• Lack of Medical Context: A general "Avoid Outdoors" warning is useless for an office commuter. Patients need advice tailored to their specific conditions (e.g., Asthma) and medications.

🟢 The Solution: RespiGuard

RespiGuard moves beyond passive monitoring to active health defense:

1. 🚦 Visual Risk Dashboard: Converts complex PM2.5 numbers into a simple Red/Yellow/Green traffic light system.
2. 🩺 AI Medical Twin: Uses RAG (Retrieval-Augmented Generation) to give advice based on your specific condition and meds.
3. 🧠 Pragmatic Risk Engine: Distinguishes between "Exercise" (Avoid) and "Commute" (Wear N95), tailored to the Indian context.
4. 🚨 Smart SOS: A "Voice for the Voiceless" panic button that broadcasts location and medical history to guardians when the patient cannot speak.

---

📸 Screenshots

Home Page	Dashboard	Ask AI
User Profile	SOS Page

---

⚙️ Technology Stack

Frontend (Client)

• React + Vite: Fast, modern UI with "Frutiger Aero" glassmorphism aesthetics.
• Tailwind CSS: For responsive and clean styling.
• Lucide React: Beautiful, consistent icons.

Backend (Server)

• Python Flask: REST API server handling business logic.
• LangChain: Orchestrator for RAG (Retrieval-Augmented Generation).
• Google Gemini 2.5 Flash: High-speed LLM for medical reasoning and chat.
• Pinecone: Vector Database for storing GINA/WHO medical guidelines.

Services & APIs

• Firebase Firestore: NoSQL database for User Profiles and History.
• OpenWeatherMap API: Real-time hyper-local AQI & PM2.5 data.
• Twilio API: SMS, WhatsApp, and Voice Call alerts for the SOS system.

---

🚀 Features

1. Hyper-Local AQI Dashboard

Real-time pollution tracking that calculates the Indian National Air Quality Index (NAQI). It provides specific "Activity Cards" telling you if it's safe to:

• 🏃 Go for a run (Outdoor Exercise)
• 🚶 Walk to work (Commute)
• 🏠 Open windows (Ventilation)

2. The AI "Medical Doctor" (Chatbot)

Ask questions like "Can I go out if I use my inhaler?"

• Context-Aware: Knows your specific condition (e.g., Bronchial Asthma) and current medications.
• Fact-Based: Answers are grounded in verified medical documents (RAG), not generic AI hallucinations.
• Memory: Remembers previous turns in the conversation for a natural flow.

3. Smart SOS System

In a respiratory emergency, every second counts. One tap triggers:

• WhatsApp Alert: Sends live Google Maps location + Patient Medical Profile to guardians.
• Voice Call: Automated call to guardians reading out the emergency context.
• First Aid Voice Mode: Reads out immediate calming steps to the patient.

---

🛠️ Installation & Setup

Prerequisites

• Node.js & npm
• Python 3.9+
• Firebase Project Credentials
• API Keys (Google Gemini, OpenWeather, Pinecone, Twilio)

1. Clone the Repository

git clone https://github.com/tulu-g559/Respi_Guard.git
cd RespiGuard

2. Backend Setup

cd server  
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate  
pip install -r requirements.txt   `

Create a .env file in /server with your keys:

GOOGLE_API_KEY=your_key  

PINECONE_API_KEY=your_key 
PINECONE_INDEX=your_index_name 
OPENWEATHER_API_KEY=your_key  

TWILIO_SID=your_sid  
TWILIO_AUTH_TOKEN=your_token  
TWILIO_FROM_NUMBER=your_number

3. Frontend Setup

cd client  
npm install

Create a .env file in /client with your Firebase config:

VITE_FIREBASE_API_KEY=your_firebase_api_key
VITE_FIREBASE_AUTH_DOMAIN=your_firebase_auth_domain
VITE_FIREBASE_PROJECT_ID=your_firebase_project_id
VITE_FIREBASE_STORAGE_BUCKET=your_firebase_storage_bucket
VITE_FIREBASE_MESSAGING_SENDER_ID=your_firebase_messaging_sender_id
VITE_FIREBASE_APP_ID=your_firebase_app_id
VITE_FIREBASE_MEASUREMENT_ID=your_firebase_measurement_id

4. Run the App

Terminal 1 (Backend):

python app.py

Terminal 2 (Frontend):

npm run dev

🔮 Future Roadmap

• Wearable Integration: Connect with smartwatches to detect SpO2 drops automatically.

• Community Heatmap: Crowdsource air quality data to find "Clean Air Pockets" in cities.

• Vernacular Voice Mode: Support for Hindi/Bengali voice commands for rural accessibility.

🤝 Contributing

Contributions are welcome! Please fork the repository and submit a pull request.