Universal AI CharterLumen Diagnose
ModelHigh riskSample dataLumen Health Labs (Company) · United States · v3.2.0
- Image Understanding
- Text Generation
Data & Training Bias
Unintended Consequences
Data-handling & compliance
- HIPAA (PHI)
- Sensitive PII
- SOC 2
Approved operating regions
- North America (NA)
NIST AI RMF functions
- Govern
- Measure
- Manage
Decision categorization (yearly disclosure)
- Healthcare › Diagnosis Support
- Healthcare › Risk Triage
- Training parameter scale
- 24B
- Fingerprinted output
- No
- Agentic decision-making
- No
- Issued
- 2026-01-15
- Effective date
- 2026-02-01
- End date
- Open-ended
- Energy / compute
- 0.6 kWh per 1M tokens
- Next disclosure due
- 2027-02-01
- Training sources
- Public clinical trial registries, de-identified radiology datasets licensed from partner hospitals, peer-reviewed oncology literature.
- Unintended consequences on record
- Early 2026 audit found lower sensitivity on scans from one imaging vendor's older hardware; model updated and reviewed quarterly since.