Preterm birth (PB) for many years remains the leading cause of neonatal morbidity, disability, and perinatal mortality as a result of severe complications in newborns in the form of intraventricular haemorrhages, necrotizing enterocolitis, respiratory distress syndrome, and other pathology. The problem has a social, medical, and economic aspect that leads to significant demographic losses and causes great financial damage to the country. Currently, effective measures for the prevention of PB have been developed, but their use should be justified by definite indications to avoid unnecessary hospitalizations and medical interventions. Scientific achievements of domestic and foreign researchers in the field of predicting PB have helped to identify predictors of premature birth, but their importance in risk definition is still not precisely defined. The search for reliable and reliable forecasting methods continues relevant despite the existing simplified mathematical calculators developed by the Fetal Medicine Foundation to calculate the risk of PB. For this reason, the use of artificial intelligence (AI) technologies, including machine learning (ML), may become a promising direction for the development of predictive analytics of PB. The decision tree, naive Bayesian classifier, random forest, support vector machine, artificial neural network, and deep neural network are the most popular machine learning methods for predicting PB today. This literature review is designed to inform a wide audience of perinatal specialists about the achievements of machine learning technologies, and the prospects of artificial intelligence in predicting PB and assessing adverse perinatal outcomes to demonstrate the basic principles of AI algorithms, the stages of creating models of learning, examples of successful use of AI methods, and the limitations of their application in clinical practice.