GEOGRAPHIC RISK DISTRIBUTION
Avg risk 13.2 across 49 states
FRAUD NETWORK
Multi-NPI connections & shared addresses
ANOMALY DEEP DIVE
Per-procedure code concentration analysis
COMPARE PROVIDERS
Side-by-side risk profile comparison
OIG TIP GENERATOR
Generate pre-filled fraud tip documents
HIGH-RISK PROVIDERS
| RISK | NPI | PROVIDER | STATE | TOTAL PAID | |
|---|---|---|---|---|---|
100 | 1003460890 | BAY SHORE BRIGHTWATERS RESCUE AMBULANCE INC. | NY | $107.5M | |
100 | 1750442844 | ADVANCED REHABILITATION OF METAIRIE | LA | $52.5M | |
100 | 1013951623 | SAMUEL FINEBERG | AL | $269.2M | |
100 | 1871930826 | 1 AFFINITY HEALTHCARE SERVICES INC | TX | $204.5M | |
100 | 1841727021 | RICHARD FERNANDES ALMEIDA | WI | $86.4M | |
85 | 1417704883 | NASER UDDIN | MD | $137.2M | |
80 | 1215029269 | VICKI BAKER | WV | $213.6M | |
80 | 1184204711 | JAMES ABERCROMBIE | CO | $217.1M | |
80 | 1699719336 | MARIA GOMES | NC | $278.8M | |
80 | 1669490348 | GORDON AHLERS | VT | $85.3M | |
80 | 1093715435 | JAMES EVANSON | ME | $55.7M | |
80 | 1821321647 | BILLINGS CLINIC | MT | $66.1M | |
75 | 1861400095 | ALEXANDER MALAYEV MD PC | ME | $69.3M | |
75 | 1780382721 | ABBY PERSONAL CARE INC. | WI | $223.4M | |
70 | 1841794211 | LEILA ABAAB DRIRA | MI | $191.6M | |
70 | 1073170494 | AIDMEN MEDICAL EQUIPMENT LLC | KY | $253.1M | |
70 | 1700167830 | ALLISON ALLEN | AK | $117.0M | |
70 | 1710416870 | NARSIS AMINIAN | DE | $140.6M | |
70 | 1124738463 | MONOGRAM HEALTH PROFESSIONAL SERVICES OF SOUTH DAKOTA, PLLC | SD | $141.0M | |
70 | 1134573777 | AMANDA ADDINGTON | TN | $200.5M | |
70 | 1750502688 | ADVANCE PHYSCIAL THERAPY AND REHABILITATION | LA | $279.8M | |
70 | 1851404396 | SUSAN ALGOOD | TN | $222.2M | |
70 | 1851395610 | JACK BROCK | NM | $161.8M | |
70 | 1548823669 | KIRSTIN ACUS | OH | $52.5M | |
70 | 1437978236 | ACUTE BEHAVIORAL HEALTH OF VIRGINIA, P.C. | TN | $212.4M |
KNOWN LIMITATIONS & UNSOLVED GAPS
Transparency is essential. The following are significant limitations of this analysis tool that users must understand before drawing any conclusions from the data presented above.
T-MSIS Data Quality Issues
HIGHTMSIS-DQ
The underlying T-MSIS (Transformed Medicaid Statistical Information System) data has well-documented quality problems that vary significantly by state.
Some states submit incomplete or delayed data — CMS has flagged data quality concerns for multiple states
Spending figures may include legitimate adjustments, recoupments, or retroactive corrections that appear as anomalies
Provider taxonomy codes are self-reported and may be inaccurate or outdated
The 2018–2024 time range includes COVID-era emergency waivers that temporarily changed billing rules
Algorithmic False Positives
HIGHALG-FP
Risk scores are based on statistical deviation from peer groups, not on verified fraudulent activity. Many flagged providers may be entirely legitimate.
High-volume specialists (e.g., oncologists, transplant surgeons) will naturally score higher due to expensive treatments
Rural providers serving large geographic areas may show unusual billing patterns that are operationally justified
New providers may appear as "spike" patterns simply because they are ramping up a legitimate practice
The composite scoring weights are heuristic-based, not validated against confirmed fraud outcomes
Limited Fraud Detection Scope
MEDIUMFRD-SCOPE
This tool only detects a narrow subset of potential fraud types. Many common healthcare fraud schemes are invisible to spending-pattern analysis alone.
Cannot detect medical necessity fraud (unnecessary procedures billed correctly)
Cannot detect upcoding (billing for more expensive services than provided)
Cannot detect phantom billing (billing for services never rendered to real patients)
Cannot detect kickback schemes (financial arrangements between providers and referral sources)
Cannot detect identity theft or credential misuse without beneficiary-level data
Historical Data Only — No Real-Time Detection
MEDIUMHIST-ONLY
All analysis is based on historical spending data from 2018–2024. This tool cannot detect fraud in progress or provide real-time alerts.
The DOGE HHS dataset was released as a static snapshot — it is not continuously updated
Providers flagged here may have already been investigated, sanctioned, or cleared
New fraud schemes emerging after the dataset cutoff date are completely invisible
Time lag between fraud occurrence and data availability can be 6–18 months
Broader Vulnerabilities Not Addressed
MEDIUMVULN-GAP
Major categories of Medicaid waste and abuse fall entirely outside the scope of provider spending analysis.
Managed care organization (MCO) fraud — capitation payment manipulation, network adequacy issues
Eligibility fraud — individuals receiving benefits they don't qualify for
Pharmacy benefit fraud — drug diversion, counterfeit medications, prescription mills
Foreign-organized fraud rings that use synthetic identities and shell companies
State-level administrative waste and misallocation of federal matching funds
No Machine Learning or Advanced Analytics
LOWNO-ML
The current risk scoring uses simple heuristic rules and statistical thresholds. No machine learning models, network analysis, or predictive analytics are employed.
A production fraud detection system would use supervised ML trained on confirmed fraud cases
Social network analysis could identify collusion rings — this tool analyzes providers in isolation
Natural language processing on clinical notes could detect documentation fraud — not available here
Beneficiary-level claims data would enable much more precise anomaly detection
This tool is intended for research and transparency purposes only. It should not be used as the sole basis for any accusation, investigation, or administrative action against healthcare providers. Any findings should be verified through official CMS, OIG, or state Medicaid fraud control unit channels.