Салдан, Й. Р.     Статистичне моделювання ризику прогресування часткової атрофії зорового нерва різної етіології [] = Statistical modeling of the risk of progression of partial optic nerve atrophy of various etiologies / Й. Р. Салдан, І. В. Галінська // Архів офтальмології України. - 2019. - Т. 7, № 1. - С. 51-56. - Бібліогр.: в кінці ст. MeSH-главная: ЛОГИСТИЧЕСКИЕ МОДЕЛИ -- LOGISTIC MODELS РЕГРЕССИОННЫЙ АНАЛИЗ -- REGRESSION ANALYSIS ЗРИТЕЛЬНОГО НЕРВА АТРОФИЯ -- OPTIC ATROPHY (диагностика, эпидемиология, этиология) ФОСФЕН -- PHOSPHENES (иммунология, физиология) Аннотация: В роботі наведено дані щодо побудови й аналізу логістичних моделей регресії тяжкості перебігу часткової атрофії зорового нерва різної етіології. Розроблено математичні моделі прогнозування низької ( 0,3) гостроти зору за показниками порогу електричної чутливості за фосфеном, коефіцієнтів MD/RNFL thickness, MD/Average GCL + IPL thickness. Також розроблено математичні моделі прогнозування тяжкої стадії за показниками коефіцієнтів MD/RNFL thickness, MD/Average GCL + IPL thickness Background. Optic nerve atrophy (ONA) is a severe multifactorial ophthalmopathy caused by common diseases (75.8 %), congenital pathology (18 %) and injuries (3–6.2 %). The disease is one of the main etiopathogenetic factors of blindness, low vision and vision disability. According to the statistics of Ukraine, the frequency of ONA as the main severe nosological form of ophthalmic pathology in primary visual disability was 8.9 %. Among the etiological factors of ONA, the diseases of the central nervous system, neuroinfections (meningitis, encephalitis, basal arachnoiditis, multiple sclerosis), ischemic heart disease, hypertension, diabetes mellitus, atherosclerosis prevail. Treatment of ONA is a complex task. The ability of the nerve tissue to regenerate is very limited, and the treatment of atrophy is reduced to increased vital activity of nerve fibers that have survived. There is no doubt that the full ONA can be attributed to incurable diseases. Therefore, all therapeutic measures relate to the so-called partial optic nerve atrophy, in which the visual functions are retained to varying degrees. The causes, nature and degree of damage to the optic nerve are diverse, making it difficult to create an optimal treatment regimen and to evaluate its efficacy. According to many authors, treatment should be combined and systematic, taking into account the etiology of the disease, the duration of the pathological process, age and general condition of the patient. The purpose of the study was to create a statistical model of the risk of progression of partial optic nerve atrophy of various etiologies. Materials and methods. Under our supervision, there were 52 patients (52 eyes) with partial ONA of various etiologies. Patients in the dynamics underwent visometry, refractometry, spheroperimetry, static Humphrey automated perimetry, tonometry, biomicroscopy, ophthalmoscopy, sonography, optical coherent tomography, phosphene threshold determination, and also calculations of ratios (mean deviation (MD)/retinal nerve fiber layer (RNFL) thickness, MD/average ganglion cell layer (GCL) + inner plexiform layer (IPL)). Methods for constructing and analyzing logistic regression models were used to identify the risk factors for ineffective treatment in patients with partial ONA of various etiologies, to evaluate the direction and degree of their influence. The observation time is 1 year. Results. Model of risk prediction does not achieve the effect by visual acuity. For the analysis of factor signs associated with the risk of not achieving the effect by visual acuity, all patients were divided into two groups: low visual acuity (initial variable Y = 1, with visual acuity 0.3) and satisfactory visual acuity (output variable Y 0.3). The analysis was conducted for 10 risk factors. To identify significant risk factors for low visual acuity, selection was performed using the step-by-step rejection-addition method. As a result, three indicators approximately equal in terms of significance were selected: threshold phosphene thresholds, ?A, MD/RNFL thickness, MD/average GCL + IPL thickness. To determine the predictive qualities of each indicator and to select the optimal threshold for predicting the risk of low visual acuity, the method of constructing and analyzing the curves of the operational characteristics of the tests was used. Selection was carried out using the step-by-step method of adding attributes to determine the significant factors associated with the risk of severe partial ONA. Two indicators — MD/RNFL thickness and MD/average GCL + IPL thickness — were selected. Model for predicting the risk of severe partial ONA. To analyze the factors related to the risk of severe partial ONA, all patients were divided into two groups: severe partial ONA (initial variable Y = 1, at stages 4 and 5), and moderate partial ONA (initial variable Y = 0, at disease stages 1–3). The analysis was carried out for 9 factors: sex, age, etiology, total photoconductivity of central visual field, total photoconductivity of peripheral visual field, RNFL, ?m, MD, PSD, MD/RNFL thickness, MD/average GCL + IPL thickness. Conclusions. The article presents data on the construction and analysis of logistic regression models of the severity of partial optic nerve atrophy of various etiologies. Mathematical models for predicting low ( 0.3) visual acuity with respect to phosphene threshold have been created. Also, mathematical models were developed to predict severe stage of the disease Доп.точки доступа: Галінська, І. В. Экз-ры: