Skip to content

Discovery note — synthetic regional payer

Setting. A utilization-review nurse receives a synthetic GLP-1 start request with drug, dose, indication, BMI, eGFR, A1c, prior therapies, and contraindication fields. Today the nurse searches a policy PDF and clinical summary, then types a determination. This is an exploration informed by Daniel Yarmoluk's regulated-healthcare, prior-authorization, payer-policy and human-sign-off background; it does not describe a customer deployment. Success means less time to a reviewable draft, every criterion traceable to a declared edge or explicit gap, and zero unreviewed submissions. Time savings are a shadow-week hypothesis, not a measured result.

Public-source findings (checked 2026-10-01)

  • The 2026 FDA Wegovy label, §§2.2, 4, 5.5, gives the injection starting dose as 0.25 mg weekly for four weeks; lists personal/family MTC, MEN2, and serious semaglutide hypersensitivity as contraindications; and calls for renal monitoring when adverse reactions could cause volume depletion. It does not provide a universal eGFR authorization cutoff. The app does not invent one.
  • The HealthPartners public weight-loss-medications policy, effective March 2024, illustrates BMI plus risk-factor, lifestyle-program, visit-attestation and concurrent GLP-1 checks. This is a pattern, not this fictional payer's policy. Its actual benefit and plan terms cannot be inferred for a case.
  • The NC Medicaid GLP-1 PA form asks for baseline BMI, adult age and weight-related complications. The UHC Medicaid announcement shows that step therapy can be plan-specific. No step requirement is copied into the demo without a declared synthetic-policy edge.
  • CMS Medicare GLP-1 Bridge FAQs distinguish coverage paths by indication and plan. This demo is a synthetic regional commercial payer; CMS material informs the discovery boundary, not its eligibility rule.

Existing CKG mapping and gaps

The separate, publicly available glp1-obesity domain in ckg-mcp has ConceptID,ConceptLabel,Dependencies,TaxonomyID rows. Relevant anchors are 28 Semaglutide, 15 BMI, 19 T2DM, 23 CKD, 26 GLP-1RA, 33 Dose escalation, 41 HbA1c endpoints, 56 Pancreatitis risk, 57 Thyroid C-cell tumor risk, and 129 Prior authorization burden. For example, declared dependency 28→26 relates semaglutide to the drug class. That edge is descriptive; it cannot establish policy eligibility. Those public descriptive edges are only discovery anchors, not review-ready authorization citations. The domain CSV is not copied into or loaded by this demo.

Missing in the bundled graph: BMI 30/27 thresholds, a weight-related-comorbidity rule, start-dose schedule, attested lifestyle program, concurrent therapy exclusion, explicit MTC/MEN2/hypersensitivity screening, step sequence, payer benefit terms, and any eGFR threshold. A1c is context, not an automatic obesity-authorization criterion. The demo's separate data/synthetic-policy.json declares the narrow fictional rules it uses, each with its source pattern. Unknown criterion, missing evidence, or contradictory submitted facts yields a gap rather than a generated assertion. A reviewer must certify any real customer's graph before use.

Scope hypothesis

Start with Wegovy injection for adult weight management only. A denial is a draft recommendation for nurse review, never an adverse benefit determination. Other drugs, indications, age groups, unusual dosing, plan exclusions, and unencoded steps stop as gaps. No real PHI belongs in this demo.