completed healthcare
MedicaidGuard Deploy
Production-style Medicaid fraud inference stack with Cloud Run API + dashboard deployment, batch scoring, and model health telemetry.
Started: March 28, 2026
Completed: April 5, 2026
Evidence & Verification
Metrics and claims on this page are tied to the linked artifacts below (repository docs, experiment outputs, and deployment pages when available).
2
Cloud Run Services
4
Api Endpoints
10/10
P0 Checks
0.8379
Auprc
Tags
healthcarefraud-detectionfastapistreamlitcloud-run
Technologies
Python FastAPI XGBoost Streamlit Cloud Run
Overview
MedicaidGuard Deploy packages healthcare fraud modeling into a deployable FastAPI service and Streamlit analyst interface. It is designed for practical Cloud Run operations, including model health checks, inference metrics, and batch throughput controls.
Live Deployment
- UI: https://medicaidguard-demo-5tphgb6fsa-as.a.run.app
- API docs: https://medicaidguard-api-5tphgb6fsa-as.a.run.app/docs
- Latest ready revisions:
medicaidguard-demo-00005-c5dmedicaidguard-api-00004-5lz
- Traffic split: 100% on latest revisions
API Surface
POST /predictfor single-claim fraud risk scoringPOST /predict/batchfor bulk scoringGET /healthfor model/runtime healthGET /metricsfor aggregated inference telemetry
Validation Snapshot
Phase 3 automated checks validated all core MedicaidGuard demo scenarios (10/10), including:
- normal/suspicious/high-risk examples,
- batch run behavior,
- risk factor visualization,
- model tab consistency (AUPRC
0.8379), - inference time target checks (
<100msscenarios).
GitHub Status
- Repository: medicaidguard-deploy
- Latest
maincommit verified on April 5, 2026:f51b4e0