Fairness Certificate
Issued April 04, 2026 — Paragon v1.0
Color Health
This certifies that the AI model identified below was trained with
algorithmic fairness constraints using the Paragon bottleneck method.
R²
0.9342
Task Performance
Equity Ratio
0.9664
worst / best group
Bottleneck d
32
latent dimension
Dataset
50,000
samples · 5 groups
Sensitive Attribute Leakage
Equity Gap: 21.5ppMODERATE
Per-Group R²
Mean: 0.9274 | Range: 0.9084 – 0.9400 | Spread: 0.0316
Regulatory Compliance
EU AI Act Article 10
Data & Data Governance
COMPLIANT
EU AI Act Article 15
Accuracy & Robustness
COMPLIANT
Automated assessment. Full compliance requires legal review.
Model Details
| Model ID | 9f6428cd2bd0c9a3ac12f0ff03a405311608011e6dc26d4ce112686b06fbbbf8 |
| Sensitive Attribute | ancestry |
| Target Attribute | disease_risk |
| Leakage | 41.5% |
| Chance Baseline | 20.0% |
| Equity Gap | 21.5pp |
| Issued (UTC) | 2026-04-04 10:00 UTC |