← CoderCoach|Platform transparency

Platform transparency

Governance

How CoderCoach manages AI, coding standards, security, and validation. Written for hospital privacy officers, HIM managers, and institutional procurement teams.

AI Governance

AI disclosure

CoderCoach uses artificial intelligence to support educational guidance and learning activities. AI-generated responses are validated against applicable coding standards where available. Learners remain responsible for exercising professional judgment when applying coding standards in clinical practice.

AI model and infrastructure

  • Text generation: Anthropic Claude (claude-sonnet-4-6 / claude-haiku-4-5)
  • Embeddings: OpenAI text-embedding-3-small
  • Retrieval: Semantic + full-text search against licensed coding standard content
  • No patient data is processed — all cases are educational / fictional
  • No AI model training on user inputs

Retrieval-augmented generation (RAG)

Before generating any AI response, the system retrieves evidence from licensed ICD-10-CA, CCI, ICD-10-CM, or ICD-10/ICD-11 content using vector similarity and full-text search. The AI is grounded in this retrieved evidence and cites sources in every answer. The standards version in use is recorded alongside each AI response.

Human verification

AI coding guidance is for educational purposes only. For production coding decisions, verify against official sources: CIHI Folio, the Canadian Coding Standards, and the DAD/NACRS Abstracting Manuals for the current fiscal year. CoderCoach does not imply certification or endorsement by CIHI or any other standards body.

Standards Governance

Standards in use

StandardVersionOwnerLicence
ICD-10-CA2026CIHILicensed
CCI2026CIHILicensed
Canadian Coding Standards2022CIHILicensed
DAD Abstracting Manual2025–2026CIHILicensed
NACRS Abstracting Manual2025–2026CIHILicensed
ICD-10-CM2026CDC / CMS (US)Public domain (US)
ICD-10 WHO2019WHOWHO licence
ICD-11 MMS2025-01WHOWHO licence (applied)
CIHI non-endorsement: This product is not affiliated with, sponsored by, or endorsed by CIHI, and CIHI has not reviewed or approved its content or functionality.

Standards update process

CIHI updates ICD-10-CA and CCI annually. When new versions are released, the platform database is updated. The standards version in use is displayed in the header on all coding screens and recorded on every AI response log. Users are notified when the platform updates to a new standards release.

Security

Infrastructure

  • Backend: Render (cloud hosting)
  • Frontend: Vercel (global CDN)
  • Database: Supabase (PostgreSQL, row-level security enabled)
  • Authentication: Supabase Auth (magic-link, JWT-based)
  • All data in transit encrypted (TLS 1.2+)
  • All data at rest encrypted

Data handling

  • No real patient data is stored or processed
  • AI questions may be processed by Anthropic and OpenAI per their data policies
  • User activity is logged for learning progress tracking
  • No PHIPA-regulated personal health information is handled

Validation Approach

AI coding answers follow a retrieval-augmented generation pipeline:

  1. The user's question is analysed to identify code categories and clinical context.
  2. Evidence is retrieved from licensed coding standard content using semantic search, full-text search, and direct code lookup.
  3. The AI generates an answer grounded in the retrieved evidence — citations show which source each answer is based on.
  4. The standards version in use is stored with every AI response for audit purposes.

This approach reduces hallucination risk significantly compared to LLM-only responses and provides an auditable trail of which documentation supported each answer.

See also: Terms & Conditions · Privacy Policy · Standards Attribution