How auzzton/med-chatbot works
CareAI is a Streamlit medical chatbot with text question answering, severity cues, medical-image interpretation, and session-based history. The visible text workflow builds retrieval QA over a FAISS index and a local Llama model; image prompts are sent to Gemini. The README also documents a hospital locator, but its stated location.py is absent from the supplied tree, so its implementation and wiring cannot be represented. Source excerpts are partial, and the available tree does not establish how dataset.txt feeds the loaded index.
Main entry points: final_model.py Streamlit app
Components 15
CareAI interface
- Streamlit app
final_model.py - Session history
final_model.py
Text answers
- Medical question
final_model.py - Answer workflow
final_model.py - Retrieval QA
final_model.py - Text embeddings
final_model.py - FAISS index
final_model.py - Local Llama
final_model.py - Severity cues
final_model.py
Image analysis
- Medical image
final_model.py - Image prompts
final_model.py - Vision workflow
final_model.py - Gemini vision
final_model.py - Image descriptions
final_model.py
Other components
- Patient
Connections 14
- Patient → Streamlit app: uses
- Streamlit app → Medical question: collects
- Streamlit app → Answer workflow: invokes
- Answer workflow → Retrieval QA: queries
- Retrieval QA → FAISS index: retrieves
- Retrieval QA → Local Llama: generates
- Answer workflow → Severity cues: assigns
- Streamlit app → Medical image: accepts
- Streamlit app → Image prompts: collects
- Streamlit app → Vision workflow: invokes
- Vision workflow → Gemini vision: submits
- Vision workflow → Image descriptions: produces
- Streamlit app → Session history: stores
- Text embeddings → FAISS index: loads index