Is artificial intelligence a solution to the myopia pandemic?

With its growing prevalence in East Asia and many parts of the world, the ‘myopia pandemic’ is estimated to affect 50% (4.7 billion) of the world’s population by 2050, with 10% (1 billion) having high myopia (≤−5.00 D).14–16 This could lead to a staggering number of myopic individuals at risk of developing blinding conditions including myopic macular degeneration (MMD) and macular neovascularisation (MNV).17 However, AI research efforts in the field of refractive errors,18 particularly myopia19 are still relatively under-developed (table 1).Table 1 Summary of current Artificial Intelligence research in myopia Title (year) Population (age group) Modalities AI model Aims Use Findings Deep learning for predicting refractive error from retinal fundus images (2018)18 General population (adults) Fundal imaging DL—CNN Refractive error prediction Diagnostic/detection Mean absolute error (MAE) of 0.56–0.91 D Prediction of myopia development among Chinese school-aged children using refraction data from electronic medical records: a retrospective, multicentre machine learning study (2018)23 General population (children) Age, SE, annual progression rate ML—random forest, mixed model, generalised estimating equation High myopia over 10 years and by age 18 Prediction High myopia over up to 10 years AUC: 3 years 0.874–0.976 5 years0.847–0.921 8 years 0.802–0.886 High myopia by 18 years old AUC: 3 years 0.940–0.985 5 years 0.856–0.901 8 years 0.801–0.837 A deep learning system for identifying lattice degeneration and retinal breaks using ultra-widefield fundus images (2019)43 General population (adolescent to adults) Ultrawide fundal images DL—CNN Notable peripheral retinal lesions (lattice or breaks) Diagnostic/detection AUC 0.999 Sensitivity 98.7% Specificity 99.2% A machine learning-based algorithm used to estimate the physiological elongation of ocular axial length in myopic children (2020)44 0 to −8D myopia (children) Demographics, SE, K, WTW, CCT ML—linear regression vs SVM vs Bagged Trees AL elongation prediction Prediction Best model: robust linear regression R2 0.87, 0.003 to 0.116 mm/year Accuracy of a deep convolutional neural network in the detection of myopic macular diseases using swept-source optical coherence tomography (2020)45 Myopia vs high myopia (adults) Swept source-OCT (SS-OCT) DL—CNN Detection of myopic macular diseases (schisis, MNV) Diagnostic/detection Detection of macular lesions: AUC 0.970 Sensitivity 90.6% Specificity 94.2% Accuracy of high m