Transforming Risk Into Value & Positive Outcomes

Invent Health is a healthcare analytics company based in Pleasanton, California that provides AI- and NLP-driven risk adjustment and quality analytics to US health plans. Its Modular Analytics Platform covers Medicare Advantage, Managed Medicaid, and ACA Marketplace lines of business in one system, and pushes prioritized care and coding gaps into provider EMRs at the point of care.

Invent Health supports Medicare Advantage including dual-eligible and D-SNP populations, Managed Medicaid including California Medi-Cal, and ACA Marketplace and Commercial plans. It supports the CMS-HCC model for Medicare Part C, Part D, and ESRD, the ACA HHS-HCC model, and Medicaid CDPS+Rx.

RAF stands for Risk Adjustment Factor. It is the score CMS uses to set risk-adjusted payment for a health plan member, calculated from the member's demographics and their documented chronic conditions as mapped to Hierarchical Condition Categories. A higher RAF reflects a member expected to cost more to care for.

V28 is the CMS-HCC risk adjustment model introduced for payment year 2024 and phased in over three years. It removed a large number of diagnosis codes from payment eligibility and lowered many condition coefficients, which reduced RAF scores for plans that had relied on the codes it dropped.

RADV stands for Risk Adjustment Data Validation. It is the CMS audit that checks whether the diagnoses a Medicare Advantage plan submitted for payment are supported by the member's medical record. Diagnoses that are not supported are recouped, and CMS now extrapolates audit findings across the contract, which materially increases the financial exposure of a failed audit.

Invent Health scores submitted HCCs by the strength of their supporting documentation, flags conditions with weak MEAT support, and simulates CMS sampling exposure, so plans can find and address unsupported diagnoses before an audit rather than after a recoupment.

Yes. Invent Health is HIPAA compliant and holds a SOC 2 attestation from an independent third-party auditor.

The HCC Coder Workbench achieves 85 to 90 percent first-pass accuracy using NLP alone, and up to 98 percent when paired with a human quality-assurance overlay. Coder-only review typically achieves 75 to 80 percent.