AI Medicare/Medicaid Billing Fraud Pattern Detection Playbook

CMS Program Integrity has referred a home health agency for review after automated edits flagged a 340% spike in high-complexity evaluation and management codes across a 6-month period. The agency's 12 clinicians are billing at the 99215 level for 94% of visits — compared to a regional peer average of 18%. You have claims data, medical records for a 50-claim sample, and the agency's internal billing guidelines.

When to use this playbook

• Use this playbook when the decision looks like the situation above: CMS Program Integrity has referred a home health agency for review after automated edits flagged a 340% spike in high-complexity evaluation and management codes across a 6-month period.
• It is a fit when you have source files in hand and need a structured, reviewable analysis — not a generic chat answer about "Medicare/Medicaid Billing Fraud Pattern Detection".
• Do not use it as a substitute for licensed, legal, clinical, or authorized official judgment in the domain.

What you'll need

• Claims data extract (12-month, all providers at the agency) - Medical records for 50-claim sample - Regional peer benchmark report - Agency's internal billing and coding guidelines

Attachments: Documents (Documents)

You are a healthcare fraud analyst and clinical reviewer conducting a Medicare billing integrity review for CMS Program Integrity.  I am attaching: - Claims data extract (12-month, all providers at the agency) - Medical records for 50-claim sample - Regional peer benchmark report - Agency's internal billing and coding guidelines

Work only from the attached source files. If a conclusion is not supported, say so.

Produce:
1. Analyze the claims data for upcoding patterns — calculate each clinician's E&M code distribution and compare against regional peer benchmarks, identifying statistical outliers by provider, code level, and service date.
2. For the 50-claim sample, cross-reference the billed code level against the medical record documentation — identify all claims where the documentation does not support the billed complexity level under CMS 1995 or 1997 guidelines.
3. Assess whether the billing pattern is consistent with documentation-driven upcoding (inadequate records written to justify higher codes) or with clinical-driven upcoding (records appear legitimate but coding is inflated).
4. Review the agency's internal billing guidelines for language that explicitly encourages high-level code assignment, targets, or incentive structures tied to billing levels.
5. Calculate the estimated improper payment amount for the full 12-month period based on the sample error rate and produce a referral package for the OIG.

Call out where independent models are likely to disagree, and list follow-up documents a reviewer should request.

What to expect

• Multi-model consensus on upcoding risk classification per provider
• E&M distribution analysis vs. peer benchmarks with outlier flags
• Documentation-to-code alignment scoring for 50-claim sample
• Billing guideline red flag register
• Estimated improper payment extrapolation and OIG referral package draft

Review before you act

• Validate this output against source files before relying on it: Analyze the claims data for upcoding patterns — calculate each clinician's E&M code distribution and compare against regional peer benchmarks, identifying statistical outliers by provider, code level, and service date.
• Validate this output against source files before relying on it: For the 50-claim sample, cross-reference the billed code level against the medical record documentation — identify all claims where the documentation does not support the billed complexity level under CMS 1995 or 1997 guidelines.
• Validate this output against source files before relying on it: Assess whether the billing pattern is consistent with documentation-driven upcoding (inadequate records written to justify higher codes) or with clinical-driven upcoding (records appear legitimate but coding is inflated).
• Validate this output against source files before relying on it: Review the agency's internal billing guidelines for language that explicitly encourages high-level code assignment, targets, or incentive structures tied to billing levels.
• Confirm every cited figure, date, counterparty, or requirement against the attached originals — models compress and can drop a qualifier.
• Treat disagreement between models as a review item, especially on classification, materiality, and recommended next action.
• Do not authorize an operational, clinical, legal, credit, or enforcement action solely because the models agree.