PA automation, by the rules.
CMS-0057-F, state parity laws, and URAC standards are transforming utilization management. Here's how GenHealth keeps you compliant — before, during, and after the 2026 deadlines.
What's required, when.
The Interoperability and Prior Authorization final rule phases in between 2026 and 2027. GenHealth's platform meets every requirement natively.
Public performance metrics
Payers publish PA metrics annually — volume, approval rate, average decision time. Our dashboards generate the reports.
API-based PA (PARDD API)
Payers must expose FHIR-based Prior Auth Requirements, Documentation, and Decision APIs. GenHealth ships with PARDD APIs.
Shortened decision windows
72 hours for urgent, 7 days for standard. Our average is 6 minutes.
Specific denial reasons
Every denial must include a specific clinical or coverage rationale. Every GenHealth decision includes the cited policy clause.
Built-in, not bolted on.
HIPAA & HITECH
Encryption at rest and in transit. Full BAA. Audit logging on every PHI access.
SOC 2 Type II
Independently audited annually. Report available under NDA.
NCQA UM standards
Medical director oversight, inter-rater reliability, appeals process — aligned with NCQA UM 4 and UM 5.
URAC Health UM
Decision timeliness, clinical review criteria, peer review — all tracked and reportable.
State parity laws
Gold-card programs, response timeframes, specialty-specific rules — configurable by state.
No autonomous denials
AI approves. Humans deny. Every adverse determination is reviewed by a licensed clinician.
Three principles, zero shortcuts.
AI approves, humans deny.
Our models recommend approvals autonomously when policy criteria are clearly met. Every potential denial routes to a licensed clinical reviewer. This is not a regulatory workaround — it's how good UM should work.
Every decision is cited.
Determinations point to the exact policy clause, clinical criterion, or guideline triggered. Auditors and appeals processes can trace every outcome to source.
Parity between AI and human review.
Inter-rater reliability is measured continuously. When the AI and reviewer disagree, disagreements train the next model version.