55.5K
Encounters
11
API Routes
TMDL
Semantic Model
Fabric
Lakehouse Proof
End-to-end healthcare analytics on Microsoft Fabric: synthetic 55.5k encounters, FastAPI read layer, dbt star schema, Power BI semantic model as TMDL in git, Fabric lakehouse validation, and MLflow-tracked XGBoost baseline. Every headline claim maps to a file in the repo.
The Microsoft Fabric + Power BI story recruiters ask for — with dbt and API layers in the same public showroom repo.
Tech Stack
Microsoft FabricPower BIdbtFastAPIPythonTMDLDAXMLflowAzure AD
Features
dbt Star Schema
Staging → intermediate → fact + dims with domain SQL tests before BI.
TMDL in Git
Certified measures and relationships reviewed like application code.
Fabric API Proof
Lakehouse + semantic validation captured as markdown + scripted screenshot summaries.
MLflow Lineage
Training runs logged with metrics — honest AUC for synthetic demo data.
How It Works
01
Serve
FastAPI exposes curated metrics and encounter slices from the synthetic dataset.
→
02
Transform
dbt builds the star schema and clinical DQ assertions.
→
03
Model
Power BI semantic model (TMDL) defines certified DAX measures.
→
04
Fabric
Workspace lakehouse + dataset checks via service principal automation.