Skip to content
1,322,867 nurse-staffing records · CMS PBJ
fonteum
DataAPIRisk SignalsResearchCompareRequest access →

Compliance posture

Methodology · Corrections log · Editorial policy

fonteum

Product

  • Data
  • API
  • Methodology
  • Sources
  • Freshness
  • Citations

For buyers

  • AI agents
  • RAG developers
  • Compliance
  • Researchers
  • Developers

Reference

  • Compare
  • llms.txt
  • Agent card
  • Audit pack
  • Pilot intake
  • Research

Sourced from federal agencies. Fonteum, Inc., Delaware C-corp. © 2026.

Fonteum · Risk Signals

Deterministic anomaly detection on the federal provider graph

Every provider record in the Fonteum graph is continuously evaluated against 5 deterministic rules: OIG exclusion status, anomalous payment patterns, specialty mismatches between NPPES and PECOS, enrollment lapses, and claim inactivity. Each signal cites the exact source row — not an ML confidence score, a primary-source fact.

OIG ExclusionHigh PaymentSpecialty MismatchEnrollment LapseNo Recent Claims
Request access →

How it works

1
Deterministic rules, not ML

Each of the 5 rules is a pure function of federal data rows. Rule a checks the OIG LEIE exclusion file. Rule b compares Open Payments totals against specialty-level percentiles. Rule c cross-joins NPPES and PECOS taxonomy columns. Rule d reads PECOS enrollment_status and MIPS recency. Rule e checks claim activity windows. No model, no threshold tuning.

2
Source-cited evidence

Every signal carries a structured evidence JSON that cites the specific source row(s) that triggered it: LEIE row ID, Open Payments total, PECOS specialty, or enrollment status. You can independently verify any signal by looking up the cited record directly in the federal source.

3
Deterministic re-runs

Signals are recomputed nightly via an Inngest cron. Each recomputation is idempotent — the same source data always produces the same signal. If the source data changes (e.g. an exclusion is removed), the signal resolves automatically and resolved_at is set.

Live signal feed

Most recent signals across the provider graph, ordered by severity then recency.

No active risk signals.

The 5 signal types

OIG Exclusion

OIG Exclusion

A provider NPI matched a record in the OIG LEIE monthly exclusion file. Excluded providers are barred from receiving payments from federal health programs. The signal cites the LEIE row ID, exclusion type, and exclusion date.

Source: OIG LEIE — oig.hhs.gov/exclusions
OIG Exclusionless than an hour ago
NPI: 1003000100
LEIE row ID12345
TypePERMEXC
Exclusion date2024-01-15
DescriptionFraud
Dataset release2026-05-01
Source: OIG LEIE Exclusions List
High Payment

High Payment

A provider's total CMS Open Payments amount for a given year exceeds the 95th percentile for their specialty. Elevated payment totals may indicate financial relationships with device or pharmaceutical manufacturers worth scrutiny in network contracting.

Source: CMS Open Payments — openpaymentsdata.cms.gov
High Payment1d ago
NPI: 1003000126
Total payments$245,000
p95 threshold$100,000
SpecialtyOrthopedic Surgery
Source: CMS Open Payments
Specialty Mismatch

Specialty Mismatch

The primary taxonomy code in NPPES does not match the specialty registered in CMS PECOS. NPPES is self-reported; PECOS is validated at enrollment. A mismatch can indicate a stale NPPES record or an administrative discrepancy that may affect network directory accuracy.

Add risk signal coverage to your provider data pipeline

Risk signals are available via the provider NPI API at /api/risk/[npi]. Each response carries full 14-tuple provenance linking the signal to its source row.

Request access →
Source: NPPES vs. CMS PECOS cross-check
Specialty Mismatch2d ago
NPI: 1003000134
NPPES taxonomy207Q00000X — Family Medicine
PECOS specialtyInternal Medicine
Source: NPPES vs. PECOS cross-check
Enrollment Lapse

Enrollment Lapse

A provider's CMS PECOS enrollment status is inactive with no recent MIPS activity. An inactive enrollment status combined with no quality reporting activity may indicate the provider has left active Medicare participation, making inclusion in credentialing rosters inaccurate.

Source: CMS PECOS enrollment — data.cms.gov
Enrollment Lapse3d ago
NPI: 1003000142
Enrollment statusI
Recent MIPS activityNo
NoteInactive enrollment status with no MIPS activity in the last performance year
Source: CMS PECOS Enrollment
No Recent Claims

No Recent Claims

A provider with an active NPI and enrollment status has no claims activity in the most recent 12-month window. This signal will activate once the claims dataset is integrated. It is currently a stub rule that will fire against CMS Part B claims data.

Source: CMS Part B Claims (pending integration)