PharmAssistAI

📑 Table of Contents

1. 📽️ Demo Videos
2. Overview
3. 🚀 Features
4. 🏗 Architecture
5. 🔧 Tech Stack
6. 📌 Future Enhancements
7. Prerequisites
8. 🔧 Setting Up the Project
   • 🐳 Docker Compose Setup
   • ⚙️ Manual Setup

---

📽️ Demo Videos

• Project Demo – Overview of the project in action.
• OCR Testing Video – Demonstration of Gemini OCR extracting text from handwritten prescriptions.
• OCR - Previous approach using Ollama – Comparison of previous OCR approach with Ollama.

---

Overview

PharmAssistAI is an AI-powered assistant designed to automate pharmacy operations, making them faster, more efficient, and error-free.

• Prescription processing via advanced OCR and LLMs
• AI-powered medicine alternatives using vector search
• Automated invoice generation & management
• Quick stock alerts based on sales.
• Agentic AI for intelligent SQL query execution
• Home remedies chatbot for customer engagement
• Interactive dashboard with inventory insights

---

🚀 Features

• Prescription Processing – Extracts text from handwritten prescriptions using Gemini-2.0-Flash-Exp OCR.
• Alternative Medicine Suggestions – Uses ChromaDB for vector-based search to suggest similar medicines.
• Home Remedies Recommendations – Uses LLM (Gemini-2.0-Flash) AI-powered chatbot provides natural remedy suggestions from trusted sources like WHO, NIH, and Mayo Clinic.
• Inventory Management – Track medicine stock levels, visualize data, and get real-time stock alerts via dashboard.
• Agentic AI for SQL – Enables natural language interaction with the database, eliminating manual SQL queries.
• Invoice Management – Automatically generates, stores, and prints invoices, with options to delete outdated records.
• Alert System – Allows you to set custom stock thresholds and provides low-stock alerts based on medicine available quantity.
• Data Insights & Visualization – Real-time analytics on sales and stock trends.

---

🏗 Architecture

---

🔧 Tech Stack

Backend:

• FastAPI – Backend server for prescription processing and API management.
• OpenCV – Image enhancement for better OCR accuracy.
• Ollama – Llama 3.2-Vision:11B for OCR (previous approach).
• ChromaDB – Vector search for alternative medicine recommendations.
• SQLite3 – Lightweight SQL database for pharmacy inventory.
• SQLAlchemy – ORM for efficient database interactions.
• Pydantic – Schema validation for data consistency.

Frontend:

• Streamlit – Rapid prototyping and interactive UI.
• Plotly – Data visualization for inventory trends.
• Pandas – Structuring and processing data in the frontend.

Agentic AI:

• Groq [Mixtral-8x7b-32768] – AI model for intelligent SQL query execution.
• LangChain & LangGraph – Frameworks for AI-driven automation.

Vector Search & Embeddings:

• ChromaDB – Vector database for retrieving semantically similar medicines.
• Sentence-Transformers – Converts medicine data into embeddings for better search.

OCR & Image Processing:

• Gemini-2.0-Flash-Experimental – OCR model for text extraction from prescriptions.

Herbal Remedies & NLP:

• Gemini-2.0-Flash – AI model for herbal remedy suggestions.

---

📌 Future Enhancements

• OAuth Authentication – Secure user access and database modifications.
• Scalability – Migrate from SQLite3 to PostgreSQL for handling large datasets.
• Enhanced UI/UX – Migrate frontend from Streamlit to React for better control and usability.

---

Prerequisites

• Python 3.9+

---

🔧 Setting Up the Project

Below are two ways to set up the project — using Docker Compose (recommended) or manual setup.

---

🐳 1️⃣ Recommended: Docker Compose Setup

Seamlessly run the entire application stack (FastAPI + Streamlit) in isolated containers with a single command.

✅ Prerequisites

• Docker: Install Docker
• Docker Compose: Install Docker Compose

---

⚙️ Steps

1. Clone the Repository

git clone https://github.com/ritu456286/PharmAssistAI.git
cd PharmAssistAI

2. Create Environment Files

• Create a .env file inside /backend and add:

GEMINI_API_KEY= <your_gemini_api_key>
GROQ_API_KEY= <your_groq_api_key>

• Create a .env file inside /frontend and add:

BASE_URL=http://backend:8000/

3. Build & Start the Application

docker-compose up --build

4. Access the Application

• Frontend (Streamlit UI): http://localhost:8501
• Backend (FastAPI docs): http://localhost:8000/docs

---

⚙️ 2️⃣ Manual Setup (Without Docker)

⚙️ Steps

1️⃣ Clone the Repository

git clone https://github.com/ritu456286/PharmAssistAI.git
cd PharmAssistAI

NOTE : Add the resources folder in the project, with medicines_ds.csv file, for mock medicines data.

2️⃣ Backend Setup

Navigate to the backend folder:

cd backend

• Create a Python virtual environment

python -m venv venv

• Activate the virtual environment

  • Windows: venv\Scripts\activate
  • Mac/Linux: source venv/bin/activate

• Install dependencies

pip install -r requirements.txt

• Create a .env file inside /backend and add:

GEMINI_API_KEY= <your_gemini_api_key>
GROQ_API_KEY= <your_groq_api_key>

• Create an empty /src/db folder where SQLite and ChromaDB will be stored.

• Start the FastAPI server

sh run.sh

---

3️⃣ Frontend Setup

Navigate to the frontend folder:

cd frontend

• Create a Python virtual environment

python -m venv venv

• Activate the virtual environment

  • Windows: venv\Scripts\activate
  • Mac/Linux: source venv/bin/activate

• Install dependencies

pip install -r requirements.txt

• Create a .env file inside /frontend and add:

BASE_URL=<url_where_your_backend_server_is_running>

• Start the Streamlit app

streamlit run main.py

---

4️⃣ Run the Application

• Ensure both backend and frontend servers are running.
• Update the inventory with your pharmacy database.
• Explore real-time stock tracking, automated billing, AI recommendations, and more!

---