Play 46
Healthcare Clinical AI
Clinical decision-support architecture with medical coding and drug interaction patterns.
Clinical decision-support architecture pattern covering medical coding (ICD-10/CPT), clinical note summarization, drug interaction checking, patient triage, audit evidence, and human review. It is not a HIPAA/HITRUST certification or medical-device approval; real deployments require customer-specific legal, clinical, security, and data-governance validation.
Architecture Pattern
RAG-powered clinical agent: FHIR data, human-in-the-loop approval, audit logging
Azure Services
DevKit (.github Agentic OS)
- agent.md — root orchestrator with builder→reviewer→tuner handoffs
- 3 agents — Healthcare Builder (gpt-4o), Reviewer (gpt-4o-mini), Tuner (gpt-4o-mini)
- 3 skills — deploy (238 lines), evaluate (166 lines), tune (236 lines)
- 4 prompts — /deploy, /test, /review, /evaluate with agent routing
- .vscode/mcp.json — FrootAI MCP with HIPAA OpenAI + FHIR inputs + envFile
TuneKit (AI Config)
- config/openai.json — gpt-4o, temp=0.1, deterministic
- config/clinical.json — coding rules, drug interaction DB, triage criteria
- config/guardrails.json — HIPAA PII redaction, human approval gates
- evaluation/eval.py — Coding accuracy >90%, Drug interaction recall >95%
Tuning Parameters
Machine evidence
FrootAI evidence lifecycle
This is an internal evidence maturity label, not third-party certification, accreditation, legal compliance, or a production guarantee. Missing or expired evidence demotes automatically; catalog claims cannot promote a play.
This play currently has design evidence only. A runnable scenario, endpoint evaluation, and build receipts are the next contiguous gates.
Repo Intelligence
v1A no-clone, revision-pinned map for agents and humans. Observed evidence is separated from inferred flow so the output stays useful without pretending to be a full call graph.