Phase 4: Add prior-auth drafting, Billing page, and PhenoML code search

[goelalex]

Adds a payer-rule-aware prior authorization flow: a new prior-auth-draft bot generates a narrative via PhenoML fhir2summary, surfaced through DraftPriorAuthModal in the encounter Details & Billing tab and saved as an inline-base64 DocumentReference. A new patient-level Billing page (route + nav tab) lists those documents and renders their narratives, and a Maya Chen demo bundle ships with a Get Started import button. Separately, a new phenoml-code-search bot backs CPT and ICD-10 autocompletes (charge items and conditions) via PhenoML Construe, replacing the ValueSet pickers. Finally, lang2fhir-create and the new bots now require a PHENOML_BASE_URL secret instead of a hardcoded URL, and both new bots are registered in deploy-bots and example-bots.json.

🤖 Generated with Claude Code

Prompt

Phase 4 — Prior-auth draft via fhir2summary

Goal: draft a prior-authorization from payer-rule-aware templates. (You preferred fhir2summary over the agent.)

• One-time setup: create narrative template(s) via POST /fhir2summary/templateseeded with payer/procedure prior-auth structure; capture the template_id.
• Bot: use similar approach with TS SDK https://github.com/PhenoML/phenoml-ts-sdk
• UI: “Draft prior auth” action from the encounter/billing area → render the returned narrative for clinician edit → save as DocumentReference/Communication.
• Manufacture the prior auth ruleset for our patient maya chen and create a fake payor.

Verify: for a patient with a coded condition + planned procedure, generate a prior-auth narrative referencing payer rules; editable and savable.

Example prior auth python code demo: https://github.com/PhenoML/phenoml-demos/tree/main/tms-prior-auth

flowchart TD
A["Referral note / PDF"]

subgraph S1["Step 1 · Intake & evidence"]
    A --> B["FHIR transaction Bundle"]
    B --> C["IPS narrative"]
    C --> D["Referral agent reviews IPS,<br/>asks follow-ups"]
    D --> E["Write Patient + follow-up<br/>answers to the EHR"]
end

subgraph S2["Step 2 · Evaluate prior auth"]
    E --> F["BCBS-MA policy agent<br/>system prompt = Policy #297"]
    F --> G["Evaluate case against<br/>policy criteria"]
end

subgraph S3["Step 3 · Adjudicate & submit"]
    G --> H["Assemble submission<br/>+ CPT codes"]
    H --> I{"Meets Policy #297?"}
end

I -->|yes| J(["APPROVED<br/>covered codes + limits"])
I -->|no| K(["DENIED<br/>rationale + citations"])