Despite the achievements of modern science in the diagnosis of myopia, there are some difficulties in determination of the trigger chains of this disease, especially in relation to each individual case. For practical work it is very important to create a mathematical model, when at the initial reception of a patient the doctor is able to predict the risk of development of acquired myopia in a particular patient. Aim of the research – to develop a mathematical model of the individual probability of development of myopia in children. Object and methods. We examined 52 children (104 eyes) aged 6 to 13 years without ophthalmic pathology. Visual acuity in all children was 1.0. The observation period was 12-24 months. A dynamic monitoring of this group of children showed that 26 children (52 eyes) further developed myopia (I observation group) and in 26 children (52 eyes), myopia was not observed (II group of observation). In addition to the standard ophthalmologic examination we determined the presence of phenotypic signs of the connective tissue dysplasia syndrome and the degree of its severity. To construct a mathematical model of the individual prognosis of the probability of development of myopia in children we used the equation of binary logistic regression. The statistical significance of the model was calculated by the Omnibus Test. To assess the relative quality of the model, ROC analysis, as well as 95% CI, was used. Results and discussion. To select the most significant signs associated with the risk of myopia, a method of stepby-step exclusion was used, which selected 7 factorial signs with certain regression coefficients. The model of the prognosis of the probability of development of acquired myopia in children included refractive corneal force, axial length of the eye, radius of curvature of the cornea, cornea diameter, reserve of relative accommodation, degree of connective tissue dysplasia, heredity of the disease. Classification ability of the model was determined according to the data of the training sample and amounted to 84.6%. At the same time, the probability of a positive result when using this model was 80.8%, and the probability of a true negative result – 88.5%. The sensitivity of the model was equal to 87.5%, and the specificity was 82.1%. The statistical significance of this model was confirmed by the Omnibus Test (χ2 = 24,479; df = 7; p <0.001). The evaluation of the quality of the model using the ROC analysis showed a good quality of the classification of characteristics: AUC was 0.897, p <0.0001 (95% CI 0.81-0.98). Conclusion. The risk of acquired myopia was determined by a set of factors: individual anatomical-optical parameters of the eye, state of accommodation, heredity, which were closely related to the manifestations of connective tissue dysplasia. Prospects for further research are to develop measures aimed at the prevention and timely detection of children prone to the development of acquired myopia.
myopia, children, prognostication, mathematical model
«Bulletin of problems biology and medicine» Issue 1 Part 2 (143), 2018 year, 223-227 pages, index UDK 617.532-039.36:519.711-0532
