How arjun1502919/gemma-hackathon-314 works
M.E.R.C.Y is a Streamlit health-education companion with symptom triage, regional-language assistance, and image-based medical-report explanations. Users enter symptoms or a question, or upload a report; the app prepares the relevant prompt and context, calls Gemma, and presents the response with safety guidance. Triage results also update consultation and dashboard state. The README describes a reference CSV that is not present in the supplied tree, and APP2.py excerpts omit parts of the UI and handlers, so uncertain wiring is not shown.
Main entry points: gemma_client.py Gemma API client · APP2.py Streamlit application · prompts.py Safety prompts · rag.py Condition retrieval
Components 14
User Experience
- Streamlit application
APP2.py - Symptom intake
APP2.py - Language assistant
APP2.py - Report explainer
APP2.py - Health dashboard
APP2.py
Triage and Safety
- Emergency check
APP2.py - Condition retrieval
rag.py - Safety prompts
prompts.py - Triage and urgency
APP2.py
Gemma Services
- Gemma API client
gemma_client.py
Consultation State
- Consultation history
APP2.py - Doctor summary
APP2.py
Other components
- Health companion user
- Google Gemma API
Connections 16
- Health companion user → Symptom intake: enters symptoms
- Health companion user → Language assistant: asks question
- Health companion user → Report explainer: uploads report
- Symptom intake → Emergency check: checks urgency
- Streamlit application → Condition retrieval: retrieves context
- Streamlit application → Safety prompts: uses instructions
- Streamlit application → Gemma API client: requests response
- Language assistant → Safety prompts: sets language
- Report explainer → Gemma API client: sends image request
- Gemma API client → Google Gemma API: calls model
- Google Gemma API → Gemma API client: returns response
- Gemma API client → Report explainer: returns explanation
- Gemma API client → Triage and urgency: returns answer
- Triage and urgency → Consultation history: saves interaction
- Consultation history → Health dashboard: feeds metrics
- Consultation history → Doctor summary: builds summary