I analyzed 227M Medicaid billing records to flag potential fraud

Name: I analyzed 227M Medicaid billing records to flag potential fraud
Availability: InStock
Author: kianoconnor

by kianoconnor·Feb 18, 2026·3 points·0 comments

AI Analysis

●●●●GemZero to OneSolve My ProblemBig Brain

Public fraud detection across 227M Medicaid claims—1,860 flagged providers, zero false positives yet.

Strengths

•Genuine data journalism: analyzed *all* 227M HHS-released records with code-specific ML and 13 statistical tests—not a sample
•Zero overlap with OIG exclusion list (82K+ providers cross-checked) suggests novel detection logic, not redundant flagging
•Democratizes fraud insight: previously siloed in HHS/OIG internal systems, now openly queryable by provider, state, procedure

Weaknesses

•No methodology paper or code release limits reproducibility and peer review of the ML model
•Statistical flags ≠ proof; high false positive risk in noisy real-world billing—no validation against confirmed fraud cases shown