01 · Shadow Evaluation
Shadow Evaluation
Claims and clinical documents feed agents that emit AgentResult only. Orchestration writes ShadowFinding and HCCRiskProfile records with no payment impact in default mode.
Horizon Aegis · SIU · Compliance · MA Plan Finance
Horizon Aegis SIU Workbench
Horizon Aegis is continuous shadow RADV: SIU work-queue, documentation-support calibration, sample stress. Live scoring is V22/V24. V28, MEAT, and B11 stay ready, not active. Packages are CREB-RADV, AREB-RADV, PIEB-RADV, and RREB-RADV (RRADV) — profiles of the four bundles, not a fifth engine. Default mode is shadow: no payment impact until a human decision.
How The Queue Works
From Claim Signal To Sealed Package
Horizon Aegis runs continuous intelligence in shadow mode on V22/V24. Findings surface for SIU review; packaging and seal remain dual-control. V28 stays off the adjudicating path until an engagement turns that flag on.
02 · Work Queue
Work Queue Priority
Queue rank is impact × confidence × risk. SIU and compliance pull highest-priority unsupported or weak support items first; disposition stays human-owned.
03 · Seal
Seal CREB-RADV / AREB-RADV
Unsupported or weak paths draft CREB-RADV; clean supported-only paths draft AREB-RADV; plan-side provider integrity drafts PIEB-RADV; recovery-ready non-legal drafts RREB-RADV. Seal requires dual-control approval and content_hash integrity checks.
Demo Work Queue
Prioritized Findings For SIU Review
Shows fictional sample rows on this public page. Live SIU queue, calibration, and RADV simulation stay on the authenticated customer console. Decision actions remain demo-only here.
| Priority | Member | HCC | Disposition | Impact | Confidence | Package Status | Action |
|---|
Loading work queue…
Capability Status
Coded And Ready, Not Active
Three engines are in the product and held off the live adjudicating path until the corresponding CMS or court gate closes. Operators see the same roster on the authenticated console.
V28 · Ready, not active
CMS-HCC V28
Coded. Live RADV audits score V22 and V24. V28 is the payment-year transition, held until a CY2024+ engagement activates that blend.
MEAT · Ready, not active
MEAT heuristic
Coded. Disabled pending further CMS guidance. CMS has not adopted MEAT as a RADV standard. Adjudication uses ICD-10-CM Official Guidelines, AHA Coding Clinic, and MMCM Ch.7.
B11 · Ready, not active
Extrapolated reserve
Coded as a RADV-B11 scenario (t-bound, FPC). Not booked liability. The 2023 RADV Final Rule remains vacated. Sample-only exposure is the active figure.
Simulation
Sample Sizes And Ratio Extrapolation
Operators may run readiness simulations with sample sizes typically between 35 and 200 members. Ranked selection uses predicted sample probability and simulated exposure; projection may use sample-only totals or a simple ratio expansion to contract population.
Disclaimer: This is a heuristic readiness model (radv-sim), not a CMS RADV engine, actuarial twin, or recovery guarantee. Labels indicate model exposure for operational planning only.
API Reference
Operator Endpoints (horizon-aegis)
Compact reference for mesh-authenticated operators. These paths are not published on jilsovereign.com.
GET /v1/work-queue mesh / customer-console session GET /v1/foresight/signals mesh / customer-console session POST /v1/calibration/run mesh-only POST /v1/simulation/radv mesh-only POST /v1/findings/:id/decision mesh-only: accept | promote_seal POST /v1/evaluate mesh-only: shadow evaluation
The public marketing site does not proxy horizon-aegis. Seal, ingest, queue, and simulation stay mesh-authenticated on the verticals plane or behind a console session.
Next Step
Brief Your SIU And Finance Leads
Walk continuous shadow evaluation, queue prioritization, dual-control seal, and readiness simulation against your MA contract footprint.