SystemAIS-001Medical device and SaMD

Retinal Screening Assistant

Support ophthalmology screening services by detecting referable diabetic retinopathy in colour fundus images and prioritising cases for specialist review.

Device class

Class IIa (MDR)

FDA

Class II, 510(k)

EU AI Act

High-risk (Annex I)

Status: Approved for operation
State: Deployed
Owner: Cirocco Jones
Readiness: 0
Blockers: 12

A control is not effective until its effectiveness is verified and the verification is linked to evidence.

AI Risks

Risks arising from the AI System, its use cases, its models or its data.

RiskRSK-001

Missed referable disease in under-represented population

Controlled

Severity: Serious
Likelihood: Possible
Category: Performance
Residual: Accepted
Use cases: UC-001
Model versions: MV-002

Model under-performs on populations under-represented in training data. Harm: Delayed referral and avoidable vision loss.

ControlsSubgroup performance reported per release.; Monthly monitoring by site with a defined action threshold.; Clinician retains the referral decision.

VerificationDHF-2291-RMR-003Risk management report

RiskRSK-002

Automation bias in reviewing clinicians

Controlled

Severity: Serious
Likelihood: Possible
Category: Human factors
Residual: Accepted
Use cases: UC-001, UC-002
Model versions: MV-002

Reviewer accepts the recommendation without independent assessment. Harm: Incorrect referral decision.

ControlsConfidence value and image evidence shown with every recommendation.; Training record required before access.

VerificationDHF-2291-HF-001Human factors validation summary

RiskRSK-003

Performance drift after camera replacement

Assessed

Severity: Serious
Likelihood: Likely
Category: Data
Residual: Not accepted
Use cases: UC-001
Model versions: MV-002

New capture hardware shifts image characteristics away from training data. Harm: Reduced sensitivity without detection.

ControlsDrift metric on image statistics, monthly.

Verification

RiskRSK-004

Re-identification from retained images

Controlled

Severity: Minor
Likelihood: Unlikely
Category: Privacy
Residual: Accepted
Use cases: Not recorded
Model versions: Not recorded

Retained training images allow re-identification of a screening participant. Harm: Loss of confidentiality.

ControlsPseudonymisation at ingest.; Access restricted to the data platform role.

VerificationPIMS-DPIA-041Data protection impact assessment

RiskRSK-005

Progression flag applied outside intended purpose

Identified

Severity: Critical
Likelihood: Unlikely
Category: Intended purpose
Residual: Not accepted
Use cases: UC-003
Model versions: Not recorded

Draft progression use case is used clinically before authorisation. Harm: Unvalidated recall decisions.

ControlsNo control recorded.

No control is recorded, so there is nothing to verify yet.

You are acting as AI System Owner; the audit trail will say so.