Using AI to predict long-term visual acuity in patients with high myopia

Japanese researchers have developed a machine learning model that predicts the risk of visual impairment among severely short-sighted patients

A child draws with coloured pencils on a sheet of paper. She is wearing a white t shirt with blue stripes. Her pencil has white polka dots on a red background

Researchers from Tokyo Medical and Dental University have developed a machine learning model that accurately predicts long-term visual acuity in patients with high myopia.

The study, which was published in JAMA Ophthalmology, involved analysing a range of 34 variables in a group of 967 Japanese patients with high myopia. Variables included age, visual acuity and levels of myopic maculopathy.

The scientists then applied different machine learning models to the data – finding that a logistic regression-based model was the most accurate for predicting visual impairment after five years.

The researchers highlighted four factors that have the most influence over an individual’s future risk of visual impairment: baseline best corrected visual acuity, prior myopic macular neovascularization, age and category four myopic maculopathy.

Reflecting on the value of the research, the scientists shared that AI models could be used for clinical assessments of future visual acuity in myopic patients.

“Using artificial intelligence to estimate future visual acuity could help clinicians to identify and monitor patients with a high risk of vision reduction in advance,” they shared.