HealthPulse+

HACKSAGON 2026 Finalist — Built by Team Kirmada

HealthPulse is a full-stack, AI-powered healthcare platform that unifies telemedicine, an AI health assistant, pharmacy bidding, and multilingual voice interaction — all under one roof.

---

🏗️ Architecture Overview

The project is a monorepo with 4 independently running services that communicate with each other over HTTP and WebSocket:

Health_Pulse/
├── frontend/          → React + Vite UI (port 5173)
├── backend/           → Node.js + Express API (port 5000)
├── ai_agent/
│   └── medical_bot/   → Python Flask AI Chatbot & Voice Server (port 5500)
└── report_Ai/         → Python Flask OCR Report Extractor (port 5501)

---

✨ Key Features

Feature	Description
🩺 Live Telemedicine	WebRTC video calls via ZegoCloud with real-time notes & prescription sync
🤖 AI Health Assistant	RAG-grounded chatbot using Groq LLM + ChromaDB — responds in your own language
🏥 HP-ID Verified Doctors	Doctors must register with their Healthcare Professional ID
⭐ Post-Session Ratings	Patients rate doctors after each consultation; rolling average updates the doctor's profile
🎤 Multilingual Voice	Sarvam AI TTS + Whisper STT for native Hindi/English voice interactions
🧠 AI Symptom Analyzer	Describe symptoms via text or voice; LLM routes you to the right specialist
💊 Pharmacy Marketplace	Prescription is broadcast to nearby shopkeepers who bid for best price
📋 OCR Report Parser	Upload a medical report image to extract structured data via EasyOCR
🌐 Multilingual UI	Toggle between English and Hindi across the entire interface

---

🚀 Getting Started

You need 4 terminal windows open — one for each service.

Prerequisites

• Node.js v18 or higher
• Python 3.10 or higher
• MongoDB Atlas account (or local MongoDB)
• ZegoCloud account for video calls
• Groq API key for LLM

---

1️⃣ Backend (Node.js — Port 5000)

cd backend
npm install

Create a .env file inside backend/:

PORT=5000
MONGODB_URI=your_mongodb_connection_string
JWT_SECRET=your_jwt_secret_here
TWILIO_ACCOUNT_SID=your_twilio_sid
TWILIO_AUTH_TOKEN=your_twilio_auth_token
TWILIO_PHONE_NUMBER=your_twilio_phone_number
SUPABASE_URL=your_supabase_url
SUPABASE_ANON_KEY=your_supabase_anon_key

Then start the server:

node app.js
# or for auto-reload:
npx nodemon app.js

✅ Server runs at http://localhost:5000

---

2️⃣ Frontend (React + Vite — Port 5173)

cd frontend
npm install

Create a .env file inside frontend/:

VITE_ZC_SERVER_SECRET=your_zegocloud_server_secret
VITE_ZC_APP_ID=your_zegocloud_app_id
VITE_BACKEND_ROUTE=http://localhost:5000
VITE_AI_ROUTE=http://127.0.0.1:5500
VITE_VOICE_ROUTE=http://127.0.0.1:5501
VITE_SUPABASE_URL=your_supabase_url
VITE_SUPABASE_ANON_KEY=your_supabase_anon_key
VITE_BASE_URL=your_base_url

Then start the dev server:

npm run dev

✅ App runs at http://localhost:5173

---

3️⃣ AI Chatbot & Voice Server (Python Flask — Port 5500)

cd ai_agent/medical_bot

Create a .env file inside ai_agent/medical_bot/:

GROQ_API_KEY=your_groq_api_key
HUGGINGFACEHUB_API_TOKEN=your_huggingface_token

Install dependencies and run:

pip install -r requirements.txt
python app.py

✅ AI server runs at http://127.0.0.1:5500

Note for Windows users: If you see a DLL initialization failed error related to PyTorch, try running pip install torch --index-url https://download.pytorch.org/whl/cpu to install the CPU-only version.

---

4️⃣ OCR Report Extractor (Python Flask — Port 5501)

cd report_Ai
pip install -r requirements.txt
python app.py

✅ Report extraction server runs at http://localhost:5501

---

🗺️ Full Startup Checklist

Open 4 separate terminals and run in order:

#	Directory	Command	Port
1	backend/	node app.js	5000
2	frontend/	npm run dev	5173
3	ai_agent/medical_bot/	python app.py	5500
4	report_Ai/	python app.py	5501

Then open your browser to http://localhost:5173

---

🗝️ Environment Variables Summary

File	Key Variables
backend/.env	PORT, MONGODB_URI, JWT_SECRET, TWILIO_*, SUPABASE_*
frontend/.env	VITE_ZC_*, VITE_BACKEND_ROUTE, VITE_AI_ROUTE, VITE_SUPABASE_*
ai_agent/medical_bot/.env	GROQ_API_KEY, HUGGINGFACEHUB_API_TOKEN

---

🛠️ Tech Stack

Frontend: React · Vite · Tailwind CSS · Socket.IO Client · ZegoCloud WebRTC · Lucide React

Backend: Node.js · Express · MongoDB (Mongoose) · Socket.IO · JWT · bcrypt · Twilio · Supabase

AI Services: Python · Flask · ChromaDB · Groq LLM · Sarvam AI · HuggingFace Embeddings · RAG Pipeline

OCR Service: Python · Flask · EasyOCR · PyTorch · OpenCV · TheFuzz

---

👥 Authors

Team Kirmada — HACKSAGON 2026

• Yash Jangir — Full Stack & AI Lead