A generative AI chatbot leveraging clinical NLP embeddings and retrieval-augmented generation (RAG) to answer oncology-related questions with precision.
This chatbot was designed to answer queries related to HER2-positive breast cancer, grounded in the landmark publication
"Human Breast Cancer: Correlation of Relapse and Survival with Amplification of the HER-2/neu Oncogene" (Slamon et al., 1987).
It integrates unified clinical vocabulary embeddings (based on biomedical research from "Unified Clinical Vocabulary Embeddings for Advancing Precision Medicine") to improve the semantic relevance and accuracy of responses.
- 🧠 Clinical Embedding Search using cosine similarity for semantically relevant answer retrieval
- 🧾 Based on validated HER2 medical literature and clinical ontologies
- 🤖 Combines RAG pipeline with LLM generation for contextual answers
- 📈 Evaluated using BLEU, ROUGE, and manual expert review
- Transformers (GPT-Neo 125M)
- Sentence-BERT for embeddings
- Scikit-learn for similarity computation
- Fitz (PyMuPDF) for PDF parsing
- Pandas, NumPy, Pickle, Regex, Requests
- ✅ Response Accuracy (BLEU, ROUGE, expert review)
- ⚡ Real-time Response Speed
- 👍 User Satisfaction (manual testing & feedback)
User: "What does HER2-positive mean in breast cancer?"
Bot: "HER2-positive breast cancer means the cancer cells have more HER2 receptors than normal. This type of cancer tends to grow faster but responds well to targeted therapies."