Laboratory Service — AI Integration
Status: populated Owner: TBD Last updated: 2026-04-18 Companion: Service Template · 03 platform-services · 02 DDD
1. Current AI Usage
The laboratory-service does not directly invoke any AI/ML model in its baseline implementation. No vendor AI SDKs are used in this service.
Per platform policy (MODULE_SHARED_STANDARDS §13 and AI_PLATFORM §11), any future assistive AI features must route through ai-gateway-service via Kong /v1/ai/* and are subject to ABAC, HITL, and provenance requirements.
2. Planned / Future AI Features (Tier B — Assistive Only)
| Feature | Purpose | Prompt template | HITL required | Status |
|---|---|---|---|---|
| Abnormal result interpretation assist | Suggest plain-language interpretation of flagged results for clinician review | lab.result.interpret.v1 | Yes — clinician must accept before any note is written | Not implemented |
| Critical value auto-draft notification | Draft a critical value notification message for clinician review | lab.critical.draft-notify.v1 | Yes — clinician must review before dispatch | Not implemented |
| Reference range gap detection | Flag when a test result has no matching reference range in catalog | Rule-based (not LLM) | No | Not implemented |
3. Moderation Policy
- All AI-assisted outputs are read-only suggestions — they are never persisted to the clinical record without explicit human acceptance.
- PHI included in prompts is subject to minimum-necessary scoping enforced by ai-gateway-service.
- Every AI request generates an
AIProvenancerecord linked to the resource that was accepted.
4. Rationale for No Current AI
Laboratory result verification is a regulated clinical process (ISO 15189). Automated decision-making on result accuracy is out of scope for the baseline. AI may augment the clinician review workflow in future iterations, not replace the verification step.