Medical News Observer

Key points

• The study analyzed 1616 eyes from 967 patients, using 34 variables to develop machine-learning models for predicting long-term visual acuity in high myopia cases.
• Support vector machines best predicted visual acuity at 3 years, while random forests were most effective at 5 years; logistic regression excelled in assessing visual impairment risk.
• Key predictive factors crucial for evaluating future vision risks included baseline best corrected visual acuity, prior myopic macular neovascularization, age, and severe myopic maculopathy.

Researchers from Tokyo Medical and Dental University have analyzed and developed a machine-learning models to predict long-term visual acuity in patients with high myopia. High myopia, characterized by the ability to see nearby objects clearly (severe shortsightedness) but difficulty focusing on distant ones, can lead to pathologic myopia, a condition that frequently results in blindness.

The retrospective, single-center cohort study analyzed 1616 eyes from 967 patients, gathering 34 variables from medical records, including general and basic ophthalmic information, and myopic maculopathy categories based on fundus and optical coherence tomography images. 

Results

Results indicated that support vector machines were most effective in predicting best corrected visual acuity at 3 years, while random forest models excelled at 5 years. For predicting the risk of visual impairment at 5 years, logistic regression showed the best performance. The team used a nomogram to represent the model’s output in a visual and accessible way for patients and clinicians, assessing the risk of future visual impairment.

Key predictive variables identified included baseline best corrected visual acuity, prior myopic macular neovascularization, age, and category 4 myopic maculopathy, all associated with an increased risk of visual impairment at five years. Long-term visual acuity prediction models can be developed for highly myopic eyes using clinical and imaging data, improving clinical assessments of future visual acuity.

Reference

Wang, Yining, Ran Du, Shiqi Xie, Changyu Chen, Hongshuang Lu, Jianping Xiong, Daniel S. W. Ting, Kengo Uramoto, Koju Kamoi, and Kyoko Ohno-Matsui. 2023. “Machine Learning Models for Predicting Long-Term Visual Acuity in Highly Myopic Eyes.” JAMA Ophthalmology, October. https://doi.org/10.1001/jamaophthalmol.2023.4786