Medicare fraud detection engine validated on CMS DE-SynPUF Medicare Claims
Medicare provider aberrant billing pattern detection using peer-group z-scores, Isolation Forest, and cross-method validation. Built on CMS DE-SynPUF. Snowflake + SAS + Python.
Medicare fraud detection engine validated on CMS DE-SynPUF Medicare Claims
Five-notebook end-to-end pipeline for ensemble anomaly detection on CMS Medicare synthetic claims — IF, LOF, OCSVM with consensus voting, hyperparameter tuning, and expert review. Master's thesis core project.
A hand-rolled ReAct agent that plans and executes multi-step SQL/Python analysis over 6.5M+ rows of Medicare claims data, with guardrails against hallucination, a transparent reasoning-trace UI, and a 20-question evaluation harness (100% accuracy). Built with Groq, DuckDB, and Streamlit - no LangChain.
Baseline comparison of Isolation Forest, Local Outlier Factor, and One-Class SVM for unsupervised anomaly detection on CMS Medicare synthetic claims data (~175k records).
New-user active-comparator study on Medicare claims: CABG vs PCI with propensity score weighting, matching, balance diagnostics and competing-risks survival analysis. SQL cohort construction over CMS DE-SynPUF.
To associate your repository with the medicare-claims topic, visit your repo's landing page and select "manage topics."