New Machine-Learning Model Predicts Long-Term Vision Loss in High Myopia Patients
A breakthrough study conducted by researchers from the Tokyo Medical and Dental University (TMDU) has introduced a new machine-learning model capable of predicting long-term visual impairment in patients with high myopia. High myopia, a condition characterized by extreme shortsightedness, is one of the top three causes of irreversible blindness in many regions of the world. This innovative model represents a significant step in combating the global challenge of vision loss.
The researchers at TMDU developed machine-learning models to predict visual acuity in patients with high myopia. They conducted a cohort study involving 967 patients to evaluate the effectiveness of these models. The results demonstrated that a regression model accurately predicts visual acuity at 3 and 5 years, while a binary classification model can predict and visualize the risk at 5 years for individual patients, showing promise in clinical assessment and monitoring.
Machine learning has proven to be effective in predicting outcomes for various health conditions. In this particular study, the researchers focused on predicting whether individuals with severe shortsightedness would have good or bad vision in the future. High myopia not only poses inconvenience to individuals, but it can also lead to a condition called pathologic myopia, which is a significant cause of blindness.
Lead author of the study, Yining Wang, explained, We know that machine-learning algorithms work well on tasks such as identifying changes and complications in myopia, but in this study, we wanted to investigate something different, namely how good these algorithms are at long-term predictions.
To develop the model, the team collected 34 variables commonly obtained during ophthalmic examinations, including age, current visual acuity, and corneal diameter, from the cohort study participants. They then tested various popular machine-learning models, with the logistic regression-based model performing the best in predicting visual impairment at 5 years.
However, predicting outcomes is only part of the story. The researchers also recognized the importance of presenting the model’s output in a way that is easily understandable to patients and conducive to clinical decision-making. To achieve this, they utilized a nomogram, which visualizes the classification model and assigns a length to each variable, indicating its importance in predicting visual acuity. These lengths can be converted into points that provide a final score explaining the risk of visual impairment in the future.
The implications of this research are significant, as permanent vision loss can have detrimental effects on individuals both financially and physically, resulting in a loss of independence. In 2019 alone, severe visual impairment led to a decrease in global productivity estimated at $94.5 billion. While further evaluation of the model on a wider population is necessary, this study demonstrates the potential of machine-learning models to address this pressing public health concern, benefiting both individuals and society as a whole.
Reference: Machine Learning Models for Predicting Long-Term Visual Acuity in Highly Myopic Eyes by Yining Wang, Ran Du, Shiqi Xie, Changyu Chen, Hongshuang Lu, Jianping Xiong, Daniel S. W. Ting, Kengo Uramoto, Koju Kamoi and Kyoko Ohno-Matsui, 26 October 2023, JAMA Ophthalmology. DOI: 10.1001/jamaophthalmol.2023.4786
