Artificial intelligence in predicting systemic parameters and diseases from ophthalmic imaging

AI Generated Summary*

Narrative review of machine learning applied to eye images to estimate non-ocular parameters and disease, covering literature to February 2022. Fundus photography is the most used modality, then OCT, then external eye photographs. Reported performance varies: age gave R2 of 0.74 to 0.92 internally, dropping to 0.36 to 0.63 in one study's external datasets; chronic kidney disease reached AUC 0.91 internally and 0.73 to 0.84 externally; five-year cardiovascular event prediction reached AUC 0.70; BMI and blood pressure estimates were weak. Retinal layer thickness metrics classified multiple sclerosis with AUC up to 0.99. The field is characterized as nascent, with sparse external validation, uncertain generalization across ethnic groups, focus on prevalent rather than incident disease, and unresolved cost-effectiveness and interpretability.

*This summary was generated by AI and is published unedited. Oku does not alter these summaries. It may contain errors or omissions and is provided for general informational purposes only. Oku does not guarantee its accuracy, completeness, or reliability. For authoritative information, please refer to the original, peer-reviewed article.