Clinical analytics leadership could finally trust
A regional health network needed executive dashboards and predictive models grounded in governed clinical data.
The challenge
Data lived across EHR exports, spreadsheets, and legacy warehouses. Leadership had seen flashy AI demos that never connected to operational decisions.
How we approached it
Decisions before dashboards
Started with the three decisions clinical and finance leaders needed weekly - then traced data lineage for each.
One definition of truth
Built governed pipelines and a semantic layer so metrics had one definition across teams.
Reporting people use
Delivered Power BI dashboards for executives and self-serve views for department analysts.
Models when data is ready
Added readmission-risk models only after data quality thresholds were met and clinicians signed off on use.
Outcomes
Faster executive reporting
Executive reporting consolidated from days of manual prep to same-day refresh.
Clinical workflows integrated
Predictive readmission scores integrated into care-coordination workflows with explainability notes.
Board-level clarity
Technical audit gave the board clarity on data risks that had been opaque for two quarters.
Technologies involved
- Python
- dbt
- Airflow
- Snowflake
- Power BI
- scikit-learn
In their words
“Their technical audit gave our board clarity we had been missing for two quarters.”