Inspiration
This hackathon is to support the Virtue Foundation's curated dataset managed by Databricks. Healthcare planners often rely on facility directories that contain messy, incomplete, or overstated claims. A facility may list ICU, dialysis, or maternity care, but the evidence may not confirm that the service is actually available.
We wanted to help planners answer a more useful question:
Where is verified care actually missing?
What It Does
VeriCare Map lets planners select a capability, such as ICU or dialysis, and see districts shaded by care-gap score.
Each score combines:
- Verified facility supply
- District-level demand indicators
- LLM-extracted structured facility data
- Confidence and trust flags
- Human field notes
Planners can click a district, inspect facilities, review evidence snippets, add notes, and trigger reclassification.
Why It Matters
The app does not blindly count facility claims. If a hospital claims ICU but has no ICU-bed evidence, it is flagged for review instead of counted as coverage.
Human corrections persist in Lakebase and improve future scoring.
Built With
- Databricks
- Delta tables
- Lakebase
- Streamlit
- LLM extraction
- Human-in-the-loop reclassification
Github project
https://github.com/vinayroyt/databricks-ep-hackathon-2026
Databricks App
https://vericare-map-7474645043324520.aws.databricksapps.com/
Built With
- databricks
- python
- streamlit
Log in or sign up for Devpost to join the conversation.