Clinical trust: unsigned DSM specifiers read as pending, Clozapine overview reword, faster consumer refusals

Merged
BI
1mo ago
+555-259

User description

Requested by Josh · project thread

Summary

Before: 71 DSM-5-TR specifier pages and 18 universal specifiers read "Definition reviewed by PsychSift" and showed their generated definition as clinician reviewed, although no named clinician signed them. The Clozapine overview read as if the adverse effects themselves require registration.

After: those 89 entries read "Pending qualified review" and their definitions are withheld until a named clinician signs them, the same rule the site-wide "Not signed off yet" line already uses. The Clozapine overview reads: "Antipsychotic for treatment-resistant schizophrenia. Treatment requires registration and blood monitoring through the clozapine monitoring service because of serious, potentially fatal adverse effects."

How: isSpecifierClinicianReviewed now requires isNamedPerson(reviewedBy) instead of any non-empty string. One sentence changed in data/medications-snapshot.json. The sign-off queue grows from 514 to 603 specifier rows. Each change is its own commit so either can be dropped.

Areas touched: Clinical reference content, Prototypes

Verification

  • Checks run:
    • npm run test:focused for the specifier change: 412 of 413 passed, the one failure was an outdated assertion, now fixed; the three specifier test files then passed 21 and 15 of 15.
    • npm run test:focused -- --files data/medications-snapshot.json: "Tests 825 passed (825)".
    • eslint on changed files clean, typecheck:internal pass.

Risk and rollout

  • Risk: clinical display. The change is more conservative only: fewer specifier definitions show until signed. Clinical Ask also stops treating those 89 as signed evidence (catalogue-evidence.ts reads the same helper).
  • Rollback: revert either commit.
  • Provider or production effects: None. No migration, no RAG ranking files.

Notes

  • Clinical text. Stays draft until Josh signs off the Clozapine wording.

🤖 Generated with Claude Code

https://claude.ai/code/session_016H8cG4UfVRH56goC9AVmL2

Generated by Claude Code

Summary by cubic

DSM specifiers stamped with the product name "PsychSift" instead of a named clinician now read as "Pending qualified review", and their definitions are withheld until a named clinician signs them, so the 71 DSM-5-TR specifier pages and 18 universal specifiers stamped by the 2026-09-26 reconcile no longer count as signed evidence. The review tool and sign-off queue apply the same named-clinician rule the site uses, and unsigned definition wording is no longer sent to Clinical Ask synthesis. The Clozapine overview now says treatment requires registration and blood monitoring because of serious adverse effects, and the medication import script preserves this wording on refreshes. Consumer questions the pattern guard already refuses skip the LLM classifier, keeping unsupported-route refusals under the 4 s latency gate.

  • The sign-off queue grows from 514 to 603 specifier rows; a product-stamped record now shows no reviewer fields until a real sign-off overwrites it.
  • The static manifest baseline now records those 71 specifiers as unverified (ineligible) rather than locally_reviewed, and the decision record is catalogued in the repo-awareness snapshot.
  • The classifier bypass only applies when every word in the query comes from a closed list of plainly non-clinical shopping-and-household words; anything else keeps the model in the loop.

Summary by Gitar

  • Retrieval behavior:
    • Consumer questions the pattern guard already refuses skip the LLM classifier in analyzeQueryWithClassifierFallback
  • Medication import:
    • scripts/import-medications-export.ts keeps the signed-off Clozapine overview wording on export refresh
  • Site-content manifest:
    • Baseline marks the specifier records validation_unverified and ineligible, with contentHash and digests refreshed

This will update automatically on new commits.

Summary by CodeRabbit

  • Clinical Information

    • Clarified that clozapine treatment requires registration and blood monitoring through the monitoring service because of serious, potentially fatal side effects.
    • Clinical definitions and notes are shown as reviewed evidence only after clinician review; other context remains available.
  • Review Status

    • Entries without a valid sign-off from a named clinician remain pending review. The reported pending-review count is now 585.
  • Search and Responses

    • Common non-clinical consumer questions are identified as unsupported without additional classification, while queries with psychiatric context continue through review.

CodeAnt-AI Description

Make clinical review status and medication guidance clearer, and speed up unsupported consumer-query refusals

What Changed

  • Specifier entries count as clinician-reviewed only when signed by a named person. The 89 entries attributed to “PsychSift” now show as pending, and their definitions are withheld until a clinician signs them.
  • The Clozapine overview now makes clear that treatment requires registration and blood monitoring because of serious adverse effects.
  • Everyday shopping and household questions already identified as unsupported skip the classifier and receive a refusal sooner. Questions with clinical context still reach the classifier.

Impact

✅ Unsigned specifier definitions stay hidden
✅ Clearer clozapine monitoring guidance
✅ Faster consumer-query refusals

💡 Usage Guide

Checking Your Pull Request

Every time you make a pull request, our system automatically looks through it. We check for security issues, mistakes in how you're setting up your infrastructure, and common code problems. We do this to make sure your changes are solid and won't cause any trouble later.

Talking to CodeAnt AI

Got a question or need a hand with something in your pull request? You can easily get in touch with CodeAnt AI right here. Just type the following in a comment on your pull request, and replace "Your question here" with whatever you want to ask:

@codeant-ai ask: Your question here

This lets you have a chat with CodeAnt AI about your pull request, making it easier to understand and improve your code.

Example

@codeant-ai ask: Can you suggest a safer alternative to storing this secret?

Preserve Org Learnings with CodeAnt

You can record team preferences so CodeAnt AI applies them in future reviews. Reply directly to the specific CodeAnt AI suggestion (in the same thread) and replace "Your feedback here" with your input:

@codeant-ai: Your feedback here

This helps CodeAnt AI learn and adapt to your team's coding style and standards.

Example

@codeant-ai: Do not flag unused imports.

Retrigger review

Ask CodeAnt AI to review the PR again, by typing:

@codeant-ai: review

Check Your Repository Health

To analyze the health of your code repository, visit our dashboard at https://app.codeant.ai. This tool helps you identify potential issues and areas for improvement in your codebase, ensuring your repository maintains high standards of code health.