Medical record documentation
Every submitted diagnosis must be supported by a medical record that contains a MEAT (Monitored, Evaluated, Assessed, Treated) evidence span — present, or the code is flagged for deletion.
The Risk-Adjustment Documentation Integrity Engine assembles an audit-defensible coded submission — every diagnosis either captured with a documented, MEAT-supported evidence span and submitted, or flagged for deletion with a written rationale — packaged into a per-chart RADV-ready evidence file. Checked against the letter of CMS RADV audit protocols before a specialist releases it.
Risk-bearing provider groups face a two-sided squeeze: the V28 HCC model cuts the number of risk-adjustable codes and compresses revenue, while CMS now audits all ~550 MA contracts every year — up from ~60 — with a coder corps growing from 40 to ~2,000. Every unsupported diagnosis you ever submitted is newly auditable and, under the False Claims Act, expensive.
Recent settlements make the price explicit — Kaiser ~$556M, Aetna $117.7M, Independent Health up to $98M — and the DOJ's recurring theme is damning: retrospective chart reviews that added revenue-raising codes while failing to delete unsupported ones.
The Risk-Adjustment Documentation Integrity Engine exists to close that gap with a single, exhaustive standard applied identically to every chart.
We do not summarize the law and hope. Every chart is scored against a versioned rule pack tied to the exact text of 42 CFR §422.310 and CMS RADV audit documentation standards. These are the provisions each submission is held to.
Every submitted diagnosis must be supported by a medical record that contains a MEAT (Monitored, Evaluated, Assessed, Treated) evidence span — present, or the code is flagged for deletion.
The exact text supporting (or contradicting) each diagnosis is extracted and cited. No code is submitted without a documented evidence span.
Every chart is reviewed for both adds and deletions. Unsupported codes are flagged with a written rationale; the submission reflects only defensible diagnoses.
Any added or deleted diagnosis requires attestation by the treating physician or a CDI specialist. The attestation is included in the evidence file.
All ICD-10 codes are mapped to the V28 HCC model (100% operative as of PY2026). Hierarchies and exclusions are applied deterministically.
Per-chart file includes the evidence span, coder notes, physician attestation, and a completeness checklist — ready for handoff to a CMS auditor.
AI extracts and drafts. Deterministic rules — running as code, outside the model — decide what is complete. A certified risk-adjustment coder and a physician reviewer sign every release. That order is never reversed.
Upload a sample of charts and the current coded submission. We return a free completeness read: which diagnoses are unsupported, which are missing, and which are at risk.
As your authorized clerical agent, we ingest EHR notes, labs, problem lists, and claims. The AI engine retrieves MEAT evidence spans for every diagnosis and flags contradictions.
Every diagnosis is reviewed for both adds and deletions. The AI proposes actions with evidence; a certified risk-adjustment coder adjudicates each suggestion.
Evidence spans are checked against MEAT criteria; the V28 crosswalk is applied; hierarchies and exclusions are verified; provider attestation is confirmed. Any failure blocks release.
A physician or CDI specialist reviews the add/delete decisions that require attestation and signs the evidence file. High-value or contested diagnoses route to a second reviewer.
You receive the audit-defensible coded submission and per-chart RADV-ready evidence files — ready for submission to CMS or handoff to an auditor.
The deliverable is completeness itself — every diagnosis accounted for with evidence or a written rationale. Nothing is left implicit.
The gates that decide completeness are code, not a model's opinion. A drafting error cannot slip past a statutory requirement.
We prepare documentation and run searches as your clerical agent. We never contact the patient, give legal advice, or submit codes on your behalf.
Simple, predictable, and aligned with accuracy — not a cut of any revenue uplift.
Start with a free Gap Scan. Send a sample of charts and your current coded submission and we'll return a completeness read against every CMS RADV protocol.
Documentation-completeness service · not legal advice · the provider group submits every code.