🧬 HER-2/neu Q&A Chatbot

A document-grounded chatbot that answers questions about HER-2/neu oncogene amplification and prognosis in breast cancer, using a research paper an open-source LLM (BioGPT). The chatbot supports continuous improvement via feedback logging, automated evaluation, and human-in-the-loop correction.

---

🚀 Features

• 🔍 Answers biomedical questions from a specific research paper (RAG + BioGPT)
• 📄 Built on a PDF publication from Science, 1987 (Slamon et al.)
• 👍 Accepts thumbs-up/down feedback from users
• 🧪 Logs and evaluates semantic accuracy using BERTScore and cosine similarity
• 🔁 Supports continuous improvement via labeled corrections and versioning

---

🛠️ Technologies Used

• Python, Streamlit, LangChain
• microsoft/BioGPT-Large for biomedical Q&A
• FAISS for semantic retrieval
• bert-score, sentence-transformers for evaluation
• Remote logging via .streamlit/secrets.toml

---

💬 How to Run

📍 Local

git clone https://github.com/Arjithab/Humana-QA-chatbot.git
cd Humana-QA-chatbot

python3 -m venv venv
source venv/bin/activate
pip install --upgrade pip
pip install -r requirements.txt

streamlit run run_chatbot.py