How rahinooopp/osteo_guard_ai_hackathon works
OsteoGuard AI is a Streamlit clinical decision-support app for osteoarthritis. Its main workflows are asking guideline-grounded questions and submitting clinical reports for extraction, scope checking, structured summarisation, and red-flag screening. It also offers patient-record management, risk-factor display, exercise-search suggestions, and retrieval statistics. Retrieval combines local dense and keyword search, rank fusion, and reranking; generation uses a configured external language-model provider. The graph follows README-documented capabilities and shown source references; some UI call details and unsampled implementation links are documented rather than independently verified in excerpts.
Main entry points: app.py Streamlit app · backend.py Hybrid retrieval · safety.py Red-flag screening · resources.py Exercise searches
Components 19
Clinical interface
- Streamlit app
app.py - Runtime config
config.py
Guideline evidence
- Question answering
backend.py - Hybrid retrieval
backend.py - Chroma index
data/osteoarthritis_db - Guideline corpus
data/guidelines - LLM generation
backend.py - Retrieval statistics
backend.py
Report workflow
- Report extraction
reports.py - Document scope gate
scope.py - Report summarisation
backend.py - Red-flag screening
safety.py
Clinical tools
- OA risk profile
risk.py - Exercise searches
resources.py
Local records
- Patient records UI
app.py - Assessment store
records.py
Other components
- Clinician
- LLM provider
- YouTube search
Connections 22
- Clinician → Streamlit app: uses
README.md - Streamlit app → Runtime config: checks settings
app.py - Streamlit app → Question answering: submits question
README.md - Question answering → Hybrid retrieval: retrieves evidence
README.md - Hybrid retrieval → Chroma index: queries
README.md - Hybrid retrieval → Guideline corpus: searches passages
README.md - Question answering → LLM generation: prompts with evidence
README.md - LLM generation → LLM provider: requests completion
README.md - Streamlit app → Report extraction: extracts report
README.md - Streamlit app → Document scope gate: checks document
README.md - Streamlit app → Report summarisation: requests summary
README.md - Report summarisation → LLM generation: generates summary
backend.py - Report summarisation → Hybrid retrieval: retrieves related evidence
README.md - Streamlit app → Red-flag screening: screens report
README.md - Red-flag screening → LLM generation: requests screening
safety.py - Streamlit app → OA risk profile: calculates profile
app.py - Streamlit app → Exercise searches: offers searches
app.py - Exercise searches → YouTube search: builds search links
resources.py - Streamlit app → Retrieval statistics: shows statistics
README.md - Streamlit app → Patient records UI: opens records
README.md - Patient records UI → Assessment store: reads and writes
app.py - Hybrid retrieval → LLM generation: returns ranked passages
backend.py