Inherited retinal dystrophies (IRDs) are a heterogeneous group of diseases, variable in both genotype and phenotype, and among the leading causes of blindness in the working-age population in developed countries. Despite advancements in genetic testing, obtaining a molecular diagnosis remains a lengthy and challenging process, constrained by limited resources and high costs. Artificial intelligence has been proposed as a way of shortening that path.
Where AI has shown itself useful
Reviewing the current literature on AI-driven approaches in IRDs, the authors find models that have demonstrated potential in predicting disease-causing variants, in distinguishing between phenotypically similar dystrophies, and in segmenting retinal layers. Routine clinical adoption, however, remains limited.
Future applications may extend considerably further: genetic counselling, prediction of IRD progression, evaluation of the pathogenicity of novel mutations, identification of candidates for gene therapy, and prediction of personalised outcomes after treatment.
What is holding it back
The barriers the review identifies are largely not modelling problems. They include a lack of standardised nomenclature, ethnic and regional variation, inconsistent multimodal imaging quality, data ownership concerns, clinical safety, cybersecurity, and the opaque nature of AI decision-making — the “black-box” problem. Continued research and collaboration, the authors argue, are what will move these.
Why it matters
Of the applications listed, identification of gene therapy candidates is the one most likely to interest developers directly. Recruitment is a rate-limiting step in rare IRD trials, and better prediction from phenotype to genotype shortens it. It is worth noting that most of the obstacles above are organisational rather than technical — matters of standards, data governance and imaging discipline. Those are tractable, but they are not solved by better models alone.
Reference
Kiraly P, Fischer MD. Artificial intelligence applications in inherited retinal dystrophies. Graefe’s Archive for Clinical and Experimental Ophthalmology. 2026 Feb;264(2):299–307.