Surgical Complications Benchmarking
Multi-page Streamlit dashboard benchmarking real CMS hospital surgical outcomes. Includes a calibrated synthetic patient-level layer for risk adjustment and predictive modeling.
- Streamlit
- CMS Data
- Risk Adjustment
- Python
Role
Data sourcing, statistical methodology, dashboard build
Timeframe
2025
The problem
Raw CMS complication rates make hospitals look better or worse than they are, because case mix differs wildly between facilities.
Approach
- 01
Ingested and cleaned public CMS hospital outcome files, standardizing measures across reporting periods.
- 02
Built a calibrated synthetic patient-level layer to enable risk adjustment where public data is aggregate only.
- 03
Implemented risk-adjusted comparisons with confidence intervals so differences are read against uncertainty.
- 04
Shipped a multi-page Streamlit app for hospital, measure, and peer-group exploration.
Outcome
Fair peer comparisons that separate genuine performance gaps from case-mix noise.
An interactive benchmarking tool that non-technical reviewers can navigate on their own.
A documented method for extending aggregate public data to patient-level modeling.
More case studies
SaaS Retention Warehouse
A full analytics warehouse for SaaS revenue and retention. Built a star schema in DuckDB, modeled cohort retention, MRR waterfall, NRR, and LTV, and served metrics through a FastAPI + React dashboard.
Readmit-IQ
End-to-end ML system for hospital readmission risk. PySpark medallion architecture, scikit-learn modeling with SHAP explainability, FastAPI inference, and a React frontend with RAG over PubMed via Qdrant.
Infection Control Surveillance
Streamlit dashboard surveilling real CDC/CMS hospital-acquired infection data. Features funnel-plot control charts, FDR-corrected hypothesis testing, and a predictive model.