Intake Completed
Complete
Inputs Used
Model NameThoraxNet
DeveloperMedVision Labs
Submission DateApr 16, 2026
IRB Approval✓ Provided
Bias Assessment✓ Completed
External Validation✓ Included
De-identification✓ Confirmed
Model Version
Version2.1.0
Model TypeMedical Imaging
Risk ClassClass II — Moderate
ArchitectureCNN / ResNet-50
Training Data Size84,200 CT scans
Outputs / Results
Intake IDRGS-M7X4K2-AUDIT
Record Created✓ Confirmed
Fields Validated12 / 12
Evaluation Criteria Applied
Mandatory documentation package complete (IRB, bias assessment, validation data)
Model metadata conforms to Class II submission requirements (EU MDR / FDA SaMD framework)
Clinical use case scoped to outpatient radiologist decision support — within approved deployment categories
Validation Executed
Complete
Inputs Used
Intake RecordRGS-M7X4K2-AUDIT
Validation RulesetRULOS-v3.2.1
Risk Weight ProfileClass II Standard
Domains Executed5 of 5
Outputs / Results
Data GovernancePASS · 100
Clinical ValidationPASS · 88
Fairness & BiasWARN · 72
Technical StandardsPASS · 91
Post-Market Gov.PASS · 85
Evaluation Criteria Applied
All 5 validation domains executed against RULOS-v3.2.1 ruleset — zero execution errors
Each check result captured as a structured, timestamped record with input hash and output state
Fairness domain flagged: intersectional sub-group analysis present but limited to 2 demographic axes — partial compliance
External validation dataset met minimum threshold (≥3 institutional sites, n > 10,000)
Decision Notes
Validation executed without interruption. One domain flagged at warning level (Fairness & Bias — intersectional analysis scope). This flag does not constitute a blocking failure for Class II devices under the current RULOS profile, but will be carried forward as a condition in the final decision record. All other domains passed at high confidence.
Assessment Generated
Complete
Inputs Used
Validation Results5 domain records
Scoring ProfileSCORE-CLASS-II
Weight DistributionClinical 30%, Data 25%, Bias 20%, Tech 15%, Gov 10%
Model Version
Scoring EngineRegOS-Score v1.9
AlgorithmWeighted composite (WACI)
Threshold ConfigTHRESH-II-2026Q1
Outputs / Results
Data Governance
100
Clinical Validation
88
Fairness & Bias
72
Technical Standards
91
Post-Market Gov.
85
Composite Score
88
Decision Notes
Composite score of 88 exceeds the Class II approval threshold of 80. The Fairness & Bias domain score of 72 falls below the domain-level ideal (≥80) but above the minimum acceptable floor (≥65) for Class II devices. This shortfall will be carried forward as a conditional requirement — expanded intersectional analysis due within 90 days of deployment.
Decision Recorded
Approved
Inputs Used
Assessment RecordASS-M7X4K2-001
Composite Score88 / 100
Approval Threshold≥ 80 (Class II)
Decision FrameworkDECI-SAMED-2026
Model Version
Decision EngineRegOS-Decide v2.1
Rulebook VersionRB-CLASS-II-Q1-2026
Evaluator IDEVL-0041
Evaluation Criteria Applied
Composite score (88) exceeds mandatory Class II approval threshold (80)
No domain scored below minimum acceptable floor (65) — conditional risk is acceptable
Fairness domain at 72 — approval conditioned on expanded intersectional analysis within 90 days
Quarterly performance monitoring required as standard post-market condition
Outputs / Results
DecisionApproved for Regulated Use
Decision Record IDDEC-M7X4K2-2026-001
Sealed✓ Apr 16, 2026 09:14:44 AM EDT
Tamper-Evident Hashsha256:7f3c2a9e1b04d6f8c5a2e7b3d9f1c4a8e2b5f7d3c9a1e4b6f2d8c5a7e1b3f9d4
Decision Notes
ThoraxNet v2.1.0 is approved for regulated deployment within the defined clinical scope (pulmonary nodule detection, outpatient radiologist decision support). Approval is granted subject to ongoing monitoring requirements and one conditional remediation item. This decision record is sealed, tamper-evident, and reproducible — any re-execution of the same evaluation with the same inputs will produce the same outcome under the current rulebook.