ChatDoc 🩺 — AI Medical Assistant for Patient Files

ChatDoc is an intelligent medical assistant built with Streamlit and LangChain. It allows users to upload patient medical records in PDF format and ask health-related questions. The system provides responses by analyzing the document and leveraging powerful LLMs and vector embeddings.

---

🔧 Features

• Upload and parse patient medical records (PDF)
• Automatically chunk, embed, and store documents using ObjectBox vector database
• Use GROQ's LLaMA 3.1 model for fast, intelligent responses
• Integrate Google Generative AI Embeddings for document understanding
• Ask contextual questions and receive medically-informed answers
• Shows relevant document chunks used for each answer

---

📁 Sample Patient File

A sample file is provided for testing: Sample-filled-in-MR.pdf

---

🚀 How to Run This Project

1. Clone the Repository

git clone https://github.com/your-username/chatdoc.git
cd chatdoc

---

2. Create a Virtual Environment

python -m venv chatdoc_env

Activate it:

• Windows:

  .\chatdoc_env\Scripts\activate

• macOS/Linux:

  source chatdoc_env/bin/activate

---

3. Install Dependencies

Install the required Python libraries:

pip install -r requirements.txt

---

4. Set Up API Keys

You will need:

• GROQ API key
• Google AI Studio API key

Create a .env file in the root directory and add your keys:

GROQ_API_KEY=your_groq_api_key
GOOGLE_API_KEY=your_google_api_key

---

5. Run the App

Before running the app, run the setup in the notebook (if any). Otherwise, launch the app with:

streamlit run app.py

---

🧪 How to Use

1. Upload a PDF file of a patient medical record (e.g., Sample-filled-in-MR.pdf).
2. Click "Creating Vector Store" to process the document.
3. Ask any medical question in the text box (e.g., "What medication is the patient on?").
4. Get an AI-generated answer, along with the relevant document context.

---

🧠 Tech Stack

• Streamlit for interactive UI
• LangChain for chaining LLM + document retriever
• GROQ (LLaMA 3.1) for language generation
• Google Generative AI Embeddings for document representation
• ObjectBox as a vector store
• PyPDFLoader to parse medical PDFs

---

⚠️ Disclaimer

This is an AI-powered demo and not a certified medical diagnostic tool. Please consult licensed healthcare professionals for actual diagnosis or treatment.