Inspiration
Classroom Gen-AI tasks turned into something useful: a medical Q&A bot grounded in the MedQuAD dataset of 2,479 expert answers, plus a concise-summarizer.
What it does
- Medical chatbot (Tasks 3-7): answers health questions with retrieved MedQuAD answers
- Concise summarizer (Tasks 1-2): shortens long text
How I built it
- Python, retrieval over All-2479-Answers-retrieved-from-MedQuAD.csv
- Iterative tasks 3 through 7 improving the bot; per-task requirements.txt in the repo
Challenges
Keeping medical answers faithful to sources and handling out-of-scope questions safely.
What I learned
Retrieval baselines beat fancy generation for factual medical Q&A at this scale.
Links
Built With
- chatbot
- healthcare
- natural-language-processing
- python
- retrieval
