Frequently Asked Questions
Clear answers about the research, what has been validated, what has not, and how a serious partnership would begin.
Research
RetGuard is a research-stage suite of three separate retinal-image classifiers described in a Version 1 manuscript submitted to medRxiv. Public DOI assignment is pending screening. It includes fundus models for diabetic retinopathy and glaucoma, plus an OCT B-scan model for Normal/AMD/DME research labels. It is not an integrated clinical screening product.
No. RetGuard does not currently manufacture or sell a retinal camera or OCT device and does not claim an agreement, endorsement, or affiliation with an imaging manufacturer. A future partner study would use a partner's defined device and acquisition protocol.
The manuscript studies diabetic retinopathy and glaucoma on color fundus photographs, and AMD and DME within a three-class OCT B-scan task. Hypertensive retinopathy was not part of the reported model suite.
No. RetGuard is not a medical device and is not cleared or approved for diagnosis, screening, referral, or patient management. The current evidence is retrospective research evidence.
No. The manuscript reports no prospective intent-to-screen study, clinical workflow validation, target-device study, human-factors study, or demonstrated patient-outcome benefit.
The public GitHub repository at github.com/anchor-neuro/retguard contains inference source, model cards, documentation, and tagged releases. Source version 1.1.1 uses the unchanged model archives published with version 1.0.0; release checksums support file-integrity verification.