Prediction of Myopia in Adolescents through Machine Learning Methods

Area of Science:

• Ophthalmology
• Public Health
• Computer Science

Background:

• Myopia is a significant and increasing eye disease in China, affecting younger populations.
• Poor visual habits and genetic factors are primary contributors to myopia development.

Purpose of the Study:

• To investigate factors influencing myopia incidence in adolescents using machine learning.
• To develop a predictive model for early myopia detection and prevention.

Main Methods:

• Utilized univariate and multivariate correlation analyses for feature selection.
• Employed Gradient Boosting Regression Trees (GBRT) for data imputation.
• Developed a predictive model using Support Vector Machines (SVM).
• Applied data transformation techniques to enhance model accuracy.

Main Results:

• The machine learning approach achieved reasonable performance and accuracy in predicting myopia.
• Identified key factors contributing to myopia incidence in adolescents.

Conclusions:

• The proposed machine learning model offers a viable tool for understanding and potentially mitigating adolescent myopia.
• Effective prevention strategies can be informed by identifying critical risk factors.