cms-data

Investigative data journalism: quantifying fixable waste in US healthcare, one issue at a time. Open-source analysis of CMS, OECD, and federal datasets. $428.6B in savings identified so far across 8 published issues.

Model Context Protocol (MCP) server exposing structured tools for CMS claims/quality data workflows. Currently runs in demo mode against sample data with a real, unpopulated PostgreSQL schema. 84/84 automated tests passing.

SQL + Tableau dashboard analyzing U.S. hospital quality performance using CMS data. Includes data cleaning pipeline, KPI development, geographic benchmarking, and interactive measure filtering.

XGBoost pipeline predicting hospital Excess Readmission Ratios on FY2024 CMS HRRP data. SHAP explainability, 5-fold CV (R² 0.938), and a live Streamlit dashboard for per-hospital risk assessment.

SQL + Power BI + Excel analytics system built on public CMS hospital data. Tracks 30-day readmission rates, HCAHPS patient satisfaction, and clinical quality indicators across 4,800+ U.S. hospitals — with benchmarking, trend analysis, and operational efficiency reporting across three dashboard pages.

SQL analytics + AI natural language query interface built on 9M+ real CMS Medicare billing records

Deterministic CMS POS/QIES ingestion pipeline for healthcare datasets, featuring schema validation, data‑quality profiling, reporting, diagnostics, and reproducible execution.

Exploratory analysis and predictive modeling of patient satisfaction across 4,789 U.S. hospitals using CMS HCAHPS public data — Python, Scikit-learn, Matplotlib

Medicaid drug spending analysis (2019–2023) using MySQL, Power BI, and Python — built as a healthcare data analyst portfolio project.

Python analytics project using CMS hospital quality data to clean, summarize, and visualize hospital ratings, reporting patterns, and facility characteristics.

Pharma commercial analytics on CMS Open Payments + Medicare Part D. dbt + Postgres + Streamlit + Groq LLM. Live demo linked.

How to request a computing account for a new student.

SQL analytics case study on 169,635 real CMS Medicare records — 20 queries across window functions, CTEs, and multi-table JOINs analyzing cost and readmission data across 3,000+ US hospitals.

Anomaly detection and overcharge prediction in Medicare Part-B claims using unsupervised outlier detection and supervised ML (XGBoost, Random Forest, logit regression). Analyzes 800K+ Illinois claims from CMS 2015 data.

SQL + Python analysis of 64K CMS hospital record. Identifies readmission risk drivers across conditions, states, and hospitals, with an Excel dashboard and executive summary.

Checks every line on a medical bill against real Medicare rates and CMS coding rules, flags the charges that look wrong, and drafts the dispute letter. Built with LangGraph, NVIDIA Nemotron, FastAPI and Next.js.

Turns a CMS facility ID into a formatted assessment report (PDF/Word) - Streamlit app on live CMS Provider Data API.

Azure Medallion lakehouse on 9.66M CMS Medicare provider-service records — PySpark Bronze→Silver→Gold on ADLS Gen2, with Power BI + marimo dashboards surfacing 5 hero billing insights.

Pending Medicare provider enrollment data: twice-weekly behavioral-health applicants enriched from NPPES for credentialing, medical billing, EHR, and provider-network research. Free sample; $12 full edition.

Analysis of CMS Medicare utilization, services, payments, and specialty-level cost patterns.