The dissertation is devoted to improving the efficiency of predicting dental caries in people of different ages by optimizing approaches due to improving the definition of structural and functional acid resistance of enamel and its functional component by enamel resistance test taking into account age and topographic features and the use of computer neural network technologies. A retrospective analysis of statistical reports on the work of the children's dental service of Kramatorsk, Donetsk region for the period from 2006 to 2016 established, according to WHO criteria, the average level of prevalence and intensity of dental caries among the city children. The analysis of medical and statistical forms of accounting and reporting documentation of the dental service of the city for the period from 2008 to 2017 revealed a decrease in the total number of visits to dentists by adult city residents and the number of installed fillings. Scientific data on age and topographic characteristics of structural and functional acid resistance of tooth enamel of different groups, which was determined by the enamel resistance test, have been supplemented. As a result of clinical studies it was found that in the examined ages 6-7, 12-15 and 35-44 years the index of structural and functional acid resistance of enamel significantly deteriorates in the direction from the cutting edge of the central incisor and canine, as well as the hump of the second premolar to the cervical region of these teeth, at the same time as in persons aged 55-70 years, such significant differences concerned only the indicators of structural and functional acid resistance of enamel in the cervical region of all studied teeth. The dynamics of the functional component of the structural and functional acid resistance of enamel depending on age and topographic characteristics is studied. In children aged 6-7 and 12-15 years, significant differences in the indicators of the functional component in determining it at different levels of the vestibular surface of the same tooth. In persons aged 35-44 and 55-70 years, only the indicators of the functional component determined in the cervical region differed significantly from others. The topographic features of tooth enamel restitution after dosed acid exposure in the conditions of its isolation from oral fluid have been studied. It is proved that under such conditions the restitution of enamel, which is carried out by centrifugal movement of dental fluid, which provides a functional component of its structural and functional acid resistance, in children aged 6-7 and 12-15 years and adults aged 35-44 years occurs more rapidly in in the area of the equator and cervical region of the studied teeth than in the area of the cutting edge. The possibility of predicting dental caries not only by the indicator of structural and functional acid resistance of enamel, determined by the test of enamel resistance on the upper central incisor, but also by the indicator of structural and functional acid resistance on the canine and second premolar of the upper jaw, but the efficiency of such prediction is lower. A method for predicting dental caries using a software product based on computer neural network technologies, taking into account five clinical indicators, in particular, the patient's age, the intensity of dental caries, oral hygiene, structural and functional acid resistance of tooth enamel by enamel resistance and resistance its functional components. The method allows to predict the number of carious lesions after 1 year in people of different ages with an efficiency of a total of 85,68%. The highest accuracy of the forecast neural network model demonstrates in children aged 6-7 years, in whom the efficiency of prediction was 92,82%, in children 12-15 years, this figure was 90,19%, in persons aged 35-44 years, it was somewhat worse – 68,46%, as a result of which there may be some restrictions on the use of this computer program for predicting dental caries in the elderly. Due to neural network technologies, the proposed program is capable of self-learning, therefore, the effectiveness of prediction with increasing clinical database will increase.
