💊 PillPal

PillPal is a simple, retrieval-augmented medical query assistant. It uses a local vector store (FAISS) and a language model to answer medical questions based on the content of Current Essentials of Medicine.

⚠️ Disclaimer: This tool is for informational and testing purposes only. It is not intended for clinical decision-making or patient care.

---

📖 Features

• 🔍 Context-aware QA: Retrieves relevant documents and answers based solely on available context.
• 🤖 LLM-powered reasoning: Utilizes Hugging Face-hosted models (like Mistral 7B) for natural, reliable responses.
• 📦 FAISS vector store integration for efficient similarity search.
• 🎛️ Streamlit-based interactive chat UI
• 🛑 Graceful fallback: If the system doesn't know, it responds appropriately instead of hallucinating answers.

---

📦 Tech Stack

• Python 🐍
• FAISS 📚
• LangChain 🔗
• Hugging Face Inference API 🤖
• Streamlit 🌐
• Docker 🐳

---

🚀 Getting Started

📦 Install Dependencies

pip install -r requirements.txt

🔑 Set Up Environment Variables

Create a .env file in the project root:

HF_TOKEN=your_huggingface_token_here

---

🏃‍♂️ Run the App

streamlit run pillpal_bot.py

---

🐳 Run via Docker (optional)

Build and run the Docker container:

docker build -t pillpal .
docker run -p 8501:8501 --env-file .env pillpal

---

📂 Project Structure

pillpal/
├── data/
│   └── Current Essentials of Medicine.pdf
│
├── vectorstore/
│   └── db_faiss/
│       ├── index.faiss
│       └── index.pkl
│
├── app.py
├── create_memory.py
├── connect_memory.py
├── pillpal_bot.py
│
├── Dockerfile
├── requirements.txt
├── README.md
├── .env.example
└── .gitignore

---

📬 Questions, feedback, or suggestions? Reach out at 📧 khanna.niharika09@gmail.com

---