💊 MediBot Jaipur — Local Medical AI Assistant

🤖 Ask about Indian medicines + find nearest medical stores in Jaipur — powered by local AI, zero internet needed!

Features • Demo • Installation • Architecture • Tech Stack • Datasets

---

🌟 What is MediBot Jaipur?

MediBot Jaipur is a fully offline, privacy-first Medical AI Assistant that answers questions about 253,973 Indian medicines and helps you find the nearest medical store from 383 verified Jaipur pharmacies — all running locally on your laptop using Mistral 7B.

No API keys. No internet. No data leaving your machine. Just pure AI magic. 🔥

---

✨ Features

Feature	Description
💊 253K+ Medicines	Complete A-Z Indian medicine database with prices in ₹
📍 383 Jaipur Stores	Find nearest medical stores by area name
🧠 RAG Architecture	Retrieval Augmented Generation for accurate answers
🔒 100% Offline	No internet, no API costs, no data privacy concerns
⚡ Fast Search	MiniLM embeddings search 323K chunks in milliseconds
🗺️ Google Maps Links	Direct links to stores on Google Maps
📞 Store Details	Phone, timings, delivery, ratings for each store
🎨 Beautiful UI	Dark-themed Gradio interface with quick-question buttons
🤖 Mistral 7B	Powerful 7B parameter LLM running on local GPU

---

🎬 Demo

📍 Area: Malviya Nagar
💊 Question: "medicine for fever"

MediBot →
"For fever, common medicines include:
 1. Paracetamol (500mg) - ₹15-30
 2. Dolo 650 - ₹30
 3. Calpol - ₹25

 Always consult a real doctor before use.

 🏥 Nearby Medical Stores in Malviya Nagar:

 1. Apollo Pharmacy (Malviya Nagar)
    ⭐ 3.4/5 (302 reviews)
    🕐 7AM - 11PM | ❌ No delivery
    📞 9198488707
    🗺️ maps.google.com/?q=26.86,75.81

 2. City Med House (Malviya Nagar)
    ⭐ 4.7/5 (2790 reviews)
    🕐 7AM - 9PM | 🚚 Delivery: 5km
    📞 9181639077
    🗺️ maps.google.com/?q=26.86,75.80"

---

🚀 Installation

Prerequisites

• Python 3.10+
• NVIDIA GPU (8GB+ VRAM recommended)
• Ollama installed
• 16GB+ RAM

Step 1 — Clone the repo

git clone https://github.com/Jashanjindal/MediBot.git
cd MediBot

Step 2 — Install dependencies

pip install -r requirements.txt

Step 3 — Pull Mistral model

ollama pull mistral

Step 4 — Download datasets

Download these and place in data/ folder:

• A-Z Medicines Dataset of India → A_Z_medicines_dataset_of_india.csv
• Medical Q&A Dataset → medDataset_processed.csv
• Jaipur Medical Stores Dataset → jaipur_medical_stores_ml.xlsx

Step 5 — Build vectorstore (one time only!)

# Open ingest.ipynb in Jupyter and run all cells
# This takes ~26 minutes — only needed ONCE!
jupyter notebook ingest.ipynb

Step 6 — Launch MediBot!

# Open launch.ipynb and run all 4 cells
jupyter notebook launch.ipynb
# Opens at http://127.0.0.1:7860

---

📁 Project Structure

MediBot/
│
├── 📓 launch.ipynb          ← Run this every day! (4 cells, ~2 mins)
├── 📁 setup/
│   └── ingest.ipynb         ← Run once to build vectorstore
│
├── 📁 data/
│   ├── A_Z_medicines_dataset_of_india.csv
│   ├── medDataset_processed.csv
│   └── jaipur_medical_stores_ml.xlsx
│
├── 📁 vectorstore/          ← Auto-generated (not on GitHub)
├── 📄 requirements.txt
├── 📄 .gitignore
└── 📄 README.md

---

🧠 Architecture

┌─────────────────────────────────────────────────────────┐
│              USER INPUT                                  │
│   Area: "Malviya Nagar"                                 │
│   Question: "medicine for fever"                         │
└──────────────┬──────────────────────────┬───────────────┘
               │                          │
               ▼                          ▼
┌──────────────────────┐    ┌─────────────────────────────┐
│   RAG PIPELINE       │    │   LOCATION ENGINE           │
│                      │    │                             │
│  MiniLM Embeddings   │    │  Search 383 Jaipur stores   │
│         ↓            │    │  by area name / GPS         │
│  ChromaDB Search     │    │  using GeoPy distance       │
│         ↓            │    │         ↓                   │
│  Top 8 relevant      │    │  Top 3 nearest stores       │
│  medical chunks      │    │  with full details          │
└──────────┬───────────┘    └──────────────┬──────────────┘
           │                               │
           └──────────────┬────────────────┘
                          ▼
            ┌─────────────────────────┐
            │      MISTRAL 7B         │
            │  Reads medicine context │
            │  + store information    │
            │  Generates full answer  │
            │  Running on RTX GPU     │
            └────────────┬────────────┘
                         ▼
            ┌─────────────────────────┐
            │      GRADIO UI          │
            │  Dark themed chat       │
            │  Quick question buttons │
            │  Location input box     │
            └─────────────────────────┘

---

🔧 Tech Stack

Component	Technology	Why
LLM	Mistral 7B via Ollama	Best 7B model, runs on 8GB VRAM
Embeddings	MiniLM (sentence-transformers)	50x faster than LLM embeddings
Vector DB	ChromaDB	Fast similarity search
Location	GeoPy	Distance calculation between GPS coordinates
Framework	LangChain	Connects all AI components
UI	Gradio	Beautiful dark-themed web interface
Language	Python 3.10+	ML ecosystem
GPU	NVIDIA RTX 4050	Local GPU inference

---

📊 Datasets

Dataset 1 — A-Z Medicines of India
├── Total medicines    : 253,973
├── Key columns        : name, price(₹), manufacturer, composition
└── Source             : Kaggle

Dataset 2 — Medical Q&A
├── Total Q&A pairs    : 16,407
├── Key columns        : question, answer, type
└── Source             : Kaggle

Dataset 3 — Jaipur Medical Stores
├── Total stores       : 383
├── Key columns        : name, address, locality, lat/lon,
│                        phone, timings, delivery, rating
└── Coverage           : All major Jaipur localities

Combined Knowledge Base
├── Total documents    : 270,380
├── After chunking     : 323,516 chunks
└── Embedding model    : all-MiniLM-L6-v2

---

📍 Jaipur Areas Covered

Malviya Nagar    Vaishali Nagar    C-Scheme
Mansarovar       Tonk Road         Ajmer Road
Sanganer         Jagatpura         Pratap Nagar
Civil Lines      Raja Park         Shyam Nagar
Nirman Nagar     Sodala            Vidhyadhar Nagar
...and many more!

---

💡 Sample Questions

💊 Medicine Info
→ "What is Paracetamol used for?"
→ "What medicines contain Amoxicillin?"
→ "Composition of Augmentin 625?"

💰 Pricing
→ "Price of Dolo 650 in India?"
→ "Cheapest fever medicines?"

⚠️ Side Effects
→ "Side effects of Azithromycin?"
→ "Is Ibuprofen safe for children?"

📍 With Location (Jaipur specific)
→ Area: Vaishali Nagar | "medicine for cold"
→ Area: C-Scheme | "where can I get Metformin?"
→ Area: Mansarovar | "nearest 24x7 pharmacy?"

---

🗓️ Daily Usage

# Step 1 — Start Ollama (keep terminal open)
ollama serve

# Step 2 — Open launch notebook
jupyter notebook launch.ipynb

# Step 3 — Run All Cells (Ctrl+F9)
# Takes ~2 minutes total

# Step 4 — Open browser
# Go to http://127.0.0.1:7860

---

🏆 What Makes This Special

• ✅ No API costs — completely free to run forever
• ✅ Privacy first — your medical questions never leave your laptop
• ✅ Indian medicine focused — built specifically for Indian medicines with ₹ prices
• ✅ Jaipur specific — 383 verified local pharmacies with real contact details
• ✅ Location aware — finds nearest store in your area instantly using GPS data
• ✅ Massive knowledge base — 323K chunks of medical knowledge
• ✅ Production ready UI — looks and feels like a real product
• ✅ Fully reproducible — anyone can clone and rebuild in ~30 minutes

---

⚠️ Disclaimer

MediBot Jaipur is for educational purposes only. Always consult a qualified medical professional before taking any medicine. Store information may not be 100% accurate — always verify before visiting. The creators are not responsible for any medical decisions made based on this tool.

---

👨‍💻 Author

Jashan Jindal

• B.Tech CS (AI & ML) — JECRC University, Jaipur
• GitHub: @Jashanjindal
• LinkedIn: Jashan Jindal

---

📜 License

MIT License — free to use, modify, and distribute!

---

Built with ❤️ in Jaipur, Rajasthan 🇮🇳

If this helped you, give it a ⭐ on GitHub!

"Built a complete offline Medical AI with RAG architecture, local LLM inference, and location-based pharmacy search — running 100% on my laptop!"