Forecasting the Development of Myopia in Children and Youth using Classification Mathematical Models («Decision Trees»)

Abstract

Currently, the worldwide increase in the spread of myopia in children and youth is a serious problem. With the progression of myopia, there is a possibility of complications that lead to decreased vision. That is why it is important to develop models for predicting the occurrence of myopia in order to identify risk groups. The purpose of the study is to develop predictive mathematical models of classification trees for calculating the probability of myopia in children and youth. Material and research methods. 3,599 schoolchildren and students of the city of Krasnoyarsk were studied, of which 2,038 were female from 6 to 20 years old and 1,561 were male from 6 to 21 years old. All respondents were divided into two groups – patients with myopia and a control group with emmetropia. All subjects underwent comprehensive measurement of somatometric, cephalometric and ophthalmological parameters. Research results. Two decision trees have been developed to predict the likelihood of myopia. According to the first classification tree, the highest probability (83.9 %) of developing myopia was observed in males when identifying such characteristics as: the age groups of adolescence and adolescence, the values of the Rees-Eysenck index less than 103.7 and the facial index lower than 89.9. Female individuals are at the highest risk of developing myopia, equal to 83.9 % if they match There are three conditions: being in the age categories of adolescence or adolescence, a body mass index below 20.2 and a head index less than 75.9. When studying ophthalmological parameters, it was revealed that the probability of developing myopia reaches the highest probability (93.4 %) in subjects with an eyeball length greater than or equal to 24.3 mm, a radius of curvature of the corneal plane meridian less than 8.1 mm and an accommodation volume less than 8.0 dpt. Conclusion. The use of mathematical classification models "decision trees" makes it possible to identify risk groups for predicting the occurrence of myopia and carrying out preventive measures.