PRISMA-ScR conformant
Search across MEDLINE, Embase, Web of Science, and Scopus. Dual screening at title/abstract and full-text; disagreements adjudicated by a third reviewer. Protocols pre-registered on OSF.
A living model-card registry for AI/ML in ophthalmology — peer-screened, APPRAISE-AI graded, dated.
Each subspecialty maintains its own searchable library and publishes a peer-reviewed scoping-review manuscript.
Transparent methods, structured appraisal, and clear evidence gaps — visible on every card
Search across MEDLINE, Embase, Web of Science, and Scopus. Dual screening at title/abstract and full-text; disagreements adjudicated by a third reviewer. Protocols pre-registered on OSF.
Each model is appraised in duplicate across six APPRAISE-AI subdomains. The 0–100 total maps to Low, Moderate, High, and Very High bands.
Annual full-text re-search per subspecialty; continuous submissions screened within four weeks. Cards are versioned and may be marked Updated, Deprecated, or Disputed.
Every card status change is logged with date, kind, and reviewer attribution.
Each annual subspecialty snapshot will be minted with a Zenodo DOI and archived to the Internet Archive. Cite the umbrella registry until per-subspecialty DOIs are issued.
@misc{eyemodelhub2026,
author = {Balas, M. and contributors},
title = {EyeModelHub: a living scoping
registry of AI/ML in ophthalmology},
year = {2026},
volume = {1}, number = {3},
url = {https://eyemodelhub.com}
} Per-subspecialty DOIs will be minted at each annual review release.
Submissions are screened against the pre-registered inclusion criteria within four weeks. Both research prototypes and regulator-cleared products are eligible; preprints and unpublished work are excluded. Eligible submissions enter peer appraisal in duplicate using APPRAISE-AI and are published as a versioned model card with reviewer attribution.