Methodology & AI Disclosure

All data comes from official government sources. No data is invented, estimated, or privately sourced.

Many Medicaid providers are businesses registered with the Nevada Secretary of State. We match providers to their corporate filings using the provider name from NPPES against the entity name in SOS records. This process uses:

• Exact name matching — provider name matches SOS entity name exactly
• Normalized matching — strips common business suffixes (LLC, Inc, Corp, PLLC, etc.) before comparing
• Manual research — some matches are verified by hand using public records, business websites, and state filings

Each match is assigned a verification status: verified (confirmed correct), plausible (likely correct but not manually confirmed), questionable (multiple concerning factors), or unverified. The verification status is displayed on each provider's detail page.

We cross-reference provider corporations and their officers against campaign finance contributor records from both Nevada SOS (state/local) and FEC (federal) databases. This is a multi-step process:

• Corporate matching — the provider's SOS entity name is searched against contributor records. Exact matches receive confidence 0.90–0.95; normalized matches (suffix-stripped) receive 0.75–0.80.
• Officer matching — corporate officers from SOS filings are searched against individual contributor records by last name and first name prefix. Base confidence is 0.30, increased by matching city (+0.20), state (+0.10), ZIP code (+0.15), full first name (+0.10), and contribution volume.

Only matches with confidence ≥ 0.50 are stored and displayed. Every match is presented with its confidence score so users can assess reliability.

Campaign contributions are legal political activity. Displaying them alongside Medicaid data provides transparency about potential conflicts of interest but does not imply any wrongdoing.

The OIG maintains the List of Excluded Individuals/Entities (LEIE), a federal database of providers banned from participating in Medicare and Medicaid. We match active Nevada Medicaid providers against this list by NPI number (exact 10-digit match).

A match means only that the same NPI appears in both the Medicaid payment records and the LEIE database. It does not necessarily mean the provider is currently excluded or billing improperly. NPIs can be reassigned, exclusions may have been reinstated, or data entry errors in either database can produce false matches. Report D1 on the Reports page shows all matches for transparency.

The Reports page uses statistical analysis to identify patterns that federal investigators commonly look for. These reports flag patterns, not fraud.

• E&M Upcoding (C2) — compares each provider's ratio of high-complexity office visits (99214+99215) to their specialty average. Flags providers whose ratio exceeds the specialty average by 20+ percentage points. Legitimate reasons exist for high ratios, including treating sicker-than-average patient populations.
• Procedure Concentration (C3) — identifies providers with 80%+ of billing in a single procedure code (minimum $100K total). Some specialties naturally concentrate billing in few codes (e.g., home health aides billing only personal care services).
• Excluded Provider Check (D1) — NPI-based matching against the OIG LEIE as described above.
• Referral Network (E1) — identifies billing entities that submit claims on behalf of 5+ distinct servicing providers. Legitimate group practices, staffing agencies, and hospital systems will appear here alongside potentially concerning arrangements.
• Behavioral Health (F1) — aggregates spending across behavioral and mental health specialties. Behavioral health is the #1 sector for Medicaid fraud nationally, so it receives dedicated monitoring. High spending alone is not evidence of fraud.
• Hospice Monitor (G1) — tracks all hospice providers billing Nevada Medicaid. California's crackdown on hospice fraud has led to migration into neighboring states, making monitoring important for prevention.
• Spending Outlier (A2) — plots providers on a logarithmic scatter chart of payments vs. patients, color-coded by cost per patient. Providers with disproportionately high cost per patient are highlighted but not accused.

This site uses the following automated and AI-assisted processes:

• Automated data ingestion — scripts download and parse data from government APIs and bulk data files on a regular schedule.
• Database cross-referencing — SQL queries join records across multiple databases by NPI, entity name, officer name, and geographic identifiers. This is deterministic (rule-based), not machine learning.
• Confidence scoring — contribution matches use a weighted scoring algorithm that adds points for matching city, state, ZIP, name, and contribution volume. The algorithm is rule-based with fixed weights, not a trained model.
• AI-assisted code generation — the site's analysis code, build scripts, and HTML pages were developed with the assistance of AI language models (Claude by Anthropic). All code was reviewed and tested by humans before deployment.
• Plain language descriptions — HCPCS procedure codes were translated from medical terminology into plain English using AI language models, then reviewed for accuracy.
• Statistical thresholds — report thresholds (e.g., 80% concentration, 20pp upcoding gap) were selected based on published academic research on Medicaid fraud detection and federal investigation guidelines.

No machine learning models make determinations about individual providers. All flagging is based on transparent, reproducible statistical rules applied uniformly to all providers in the dataset.

• The public HHS T-MSIS dataset is aggregated. We cannot see individual claims, diagnosis codes, patient identities, or dates of service. This limits the types of analysis possible.
• Provider-to-corporation matching relies on name similarity. Providers with common business names may be incorrectly matched. Providers operating under a DBA (doing business as) name different from their legal entity may not be matched at all.
• Campaign contribution matching by officer name can produce false positives when common names (e.g., “John Smith”) match contributors in the same city. Confidence scores reflect this uncertainty.
• The OIG LEIE includes historical exclusions. Some matched providers may have been reinstated or the exclusion may predate their Medicaid billing period.
• Data freshness varies by source. HHS data may lag 6–12 months behind current billing. NPPES addresses may be outdated if providers have not updated their registration.
• County assignment uses provider practice city from NPPES, mapped to county. Providers who practice across county lines are assigned to their registered address county only.

If you identify an error in our data or analysis, please report it so we can investigate and correct it:

[email protected]

Please include the page URL, the specific information you believe is incorrect, and any evidence supporting the correction. We review all reports and update the site where warranted.