aureumhealth
Platform Overview

The complete AI governance platform

Nine integrated modules that work together to give you visibility, control, and audit evidence for every AI asset in your enterprise.

AI Catalog

Your single source of truth for AI

Maintain a complete inventory of every AI asset across your enterprise—Epic modules, vendor solutions, internal ML models, and agentic tools.

  • Asset registration with metadata and ownership
  • Version tracking and change history
  • PHI tier classification (Tier 0–3)
  • Lifecycle state management
  • Searchable, filterable registry

AI Asset Catalog

Asset NameTypeStatusPHI TierOwner
Epic Sepsis ModelVendorProductionTier 2Clinical AI Team
Fall Risk PredictionInternal MLProductionTier 3Data Science
Scheduling AssistantAgentStagingTier 0IT Operations
Clinical Notes SummarizerLLMSandboxTier 3Clinical Informatics
Radiology Auto-ReportInternal MLProductionTier 2Radiology IT
47
Total Assets
32
In Production
18
PHI Tier 2+
8
Pending Review

Onboarding & Certification

Structured paths to production

Define and enforce certification workflows that ensure every AI asset meets your governance requirements before deployment.

  • Customizable onboarding checklists
  • Documentation requirements enforcement
  • Risk assessment workflows
  • Certification expiration and renewal
  • Evidence collection and attachment

Patient Intake Chatbot

New AI Asset Onboarding
In Progress
Onboarding Progress6 of 9 steps complete
Asset RegistrationComplete
Basic metadata, ownership, and asset type defined
PHI ClassificationComplete
Classified as Tier 2 (Limited PHI exposure)
Technical DocumentationComplete
Architecture diagram, API docs, dependencies uploaded
Risk AssessmentIn Progress
Risk questionnaire 75% complete
Security ReviewPending
Available after risk assessment
Testing & ValidationPending
Sandbox testing required
Final CertificationPending
All prior steps must be complete
Documents Required
✓ Model card uploaded
✓ Data flow diagram attached
⚠️ Vendor SOC 2 report pending
Estimated time to certification5-7 business days

Approval Workflows

Separation of duties, enforced

Route AI through security, compliance, and model risk approvals with configurable gates and mandatory sign-offs.

  • Multi-stage approval workflows
  • Role-based approval routing
  • Conditional approval logic
  • Escalation paths and deadlines
  • Approval audit trail

Clinical Notes Summarizer - Approval Pipeline

In Review
Submitted by Dr. Sarah Chen • 2 days ago
Security Review
Approved
Reviewed PHI controls, data encryption, access policies
Michael Rodriguez, CISO•Jan 11, 2026
Compliance Review
Pending
Awaiting HIPAA compliance verification and audit evidence review
Assigned to: Jennifer Walsh, Compliance Officer
Model Risk Assessment
Waiting
Validate model performance, bias testing, clinical safety
Will be assigned after compliance approval
Clinical Review
Waiting
Final clinical validation and workflow integration approval
Will be assigned after model risk approval
2 of 4 approvals complete

Environment Controls

Sandbox to production, governed

Control AI progression through environments with promotion rules, testing requirements, and rollback capabilities.

  • Environment definition (Sandbox, Staging, Prod)
  • Promotion rules and gates
  • Environment-specific configurations
  • Testing requirements before promotion
  • Environment isolation and access controls

Radiology Auto-Report Model

Environment Promotion Pipeline
Sandbox
Dev
v3.2-dev
Testing in progress
PHI Controls✓
Unit Tests✓
Code Review✓
Staging
QA
v3.1
Awaiting approval
Integration Tests✓
QA Sign-off✓
Prod Approval⏳
Production
Live
v3.0
Active - Stable
Uptime99.9%
Accuracy92.1%
Alerts0
Staging → Production Promotion Rules
✓ All integration tests passed
✓ QA team approval obtained
⏳ Security review approval (pending)
⏳ Clinical validation approval (pending)
• Performance metrics within threshold
• No critical issues in last 7 days

Install & Rollback

Operational state machine

Execute controlled deployments with instant rollback capabilities and complete operational auditability.

  • One-click install and rollback
  • Pre-deployment validation checks
  • Deployment scheduling and windows
  • Automatic rollback triggers
  • Deployment impact analysis

Fall Risk Prediction Model v2.3

Production Deployment
Active
Deployment History
v2.3 - CurrentDeployed
Deployed on Jan 14, 2026 at 2:15 PM by System
✓ Pre-deploy checks passed•Performance: 89.2%
v2.2
Deployed Jan 8, 2026 • Runtime: 6 days
v2.1Archived
Deployed Dec 20, 2025 • Runtime: 19 days
Pre-Deployment Validation
Security scan✓ Pass
Dependency check✓ Pass
Performance baseline✓ Pass
Governance approvals✓ Pass
Environment compatibility✓ Pass
Automatic Rollback Enabled
Will auto-rollback if accuracy drops below 85% for 2+ hours

Performance & Drift

Continuous governance awareness

Monitor AI performance and detect drift with thresholds tied to governance decisions and automatic alerting.

  • Real-time performance metrics
  • Drift detection algorithms
  • Governance threshold configuration
  • Automatic alert routing
  • Performance trend analysis

Fall Risk Prediction Model

Last 30 days • Production Environment
Drift Detected
Accuracy
84.2%
-3.8%
Precision
81.5%
-4.2%
Predictions/Day
1,247
+12%
Performance Trend (30 days)
30 days agoToday
Performance Below Threshold
Accuracy has dropped below the 88% governance threshold. Model may require retraining or review.

Governed Explainability

Transparency with controls

Provide tiered explainability (L1/L2/L3) with policy-controlled access to SHAP, LIME, and model internals.

  • L1: High-level factor explanations
  • L2: Feature contribution details
  • L3: Full model internals access
  • Role-based explainability access
  • Explainability audit logging

Sepsis Risk Prediction - Case #18247

Patient Age: 67 • Admitted: Jan 12, 2026
High Risk
Confidence: 89%
Level 1: Summary Explanation
Available
High risk prediction based on elevated vital signs, laboratory values, and clinical history. Key factors include white blood cell count, temperature, and recent infection history.
WBC ElevatedTemperatureInfection History
Level 2: Feature Importance
Authorized
WBC Count18.2 K/µL
+34%
Temperature38.9°C
+28%
Heart Rate112 bpm
+18%
Lactate Level2.8 mmol/L
+12%
Recent SurgeryYes (3 days ago)
+8%
Level 3: Model Internals (SHAP/LIME)
Restricted
Detailed SHAP values, model coefficients, and internal decision boundaries require Level 3 authorization.
Explanation Access Log
Dr. James Patterson accessed L1 & L2 explanations • 2 mins ago
Nurse Sarah Kim accessed L1 explanation • 15 mins ago

Usage & Cost Governance

AI spend under control

Track and allocate AI costs with showback and chargeback capabilities across teams, projects, and models.

  • Token and compute usage tracking
  • Cost allocation by dimension
  • Showback reporting dashboards
  • Chargeback integration
  • Budget alerts and limits

AI Usage & Cost Dashboard

Total Spend
$47.2K
+18% vs last month
LLM Tokens
847M
$31.2K cost
ML Compute
1,243h
$12.8K cost
API Calls
2.4M
$3.2K cost
Cost by Department
Clinical Operations
$18.4K / $20K92%
Radiology
$12.8K / $12K107%
Emergency Dept
$9.2K / $10K92%
Research & Analytics
$6.8K / $8K85%
Top 5 Models by Cost
Clinical Notes Summarizer
$12,340
234M tokens
Radiology Auto-Report
$8,920
847 hrs compute
Scheduling Assistant
$7,650
892K calls
Fall Risk Prediction
$5,240
421 hrs compute
Sepsis Detection
$4,180
156M tokens
Budget Alert: Radiology
Department has exceeded monthly budget by 7%. Consider reviewing usage or increasing allocation.

Immutable Audit Logs

Compliance-ready evidence

Maintain append-only audit trails for every action, decision, and state change—ready for regulators and auditors.

  • Append-only log architecture
  • Tamper-evident record chain
  • Configurable retention policies
  • Export for audit submissions
  • Real-time audit dashboards

Immutable Audit Trail

Cryptographically verified • All entries are tamper-evident and append-only
Accessed L2 Explainability
Sepsis Risk Model
Explainability Access
Dr. Michael Chen•2026-01-14 14:23:18 UTC
Approved for Production
Clinical Notes Summarizer
Approval
Sarah Walsh, Compliance•2026-01-14 14:18:42 UTC
Performance Alert Triggered
Fall Risk Prediction
Monitoring
System•2026-01-14 14:12:05 UTC
Deployed to Staging
Scheduling Assistant v2.1
Deployment
James Rodriguez, IT•2026-01-14 13:45:33 UTC
Updated Model Parameters
Radiology Auto-Report
Configuration Change
Emily Park, Data Science•2026-01-14 13:22:17 UTC
Security Review Approved
Patient Intake Chatbot
Approval
Michael Rodriguez, CISO•2026-01-14 12:58:09 UTC
247,382
Total Events
847
Today
100%
Verified

See the platform in action

Schedule a Demo