The use of artificial intelligence to diagnose diseases and predict their outcomes in newborns

Abstract

   In recent years, modern models of artificial intelligence, including neural networks, have been successfully introduced into clinical practice, due to the high accuracy of functioning and the prospects of their use for the diagnosis and prediction of various diseases.   Purpose. To improve the processes of predicting and diagnosing diseases and their outcomes in newborns using neural network intelligent technologies.   Material and methods. The study is based on statistically reliable collection of patient history data, mathematical analysis, fuzzy logic theory and principles of trainable neural network systems.   Results. Neural network programs have been developed to predict the course of posthypoxic disorders of the cardiovascular system in newborns; to determine the probability of occurrence and outcomes in newborns of such significant diseases as cerebral leukomalacia, intracranial hemorrhages, hydrocephalus, necrotizing enterocolitis, bronchopulmonary dysplasia, retinopathy of prematurity, early anemia of prematurity; to predict the physical and neuropsychiatric development of a child to age of one year; and also to predict an unfavorable outcome (death or disability with persistent health problems) of children born earlier than 32 weeks of gestation.   Conclusion. The developed artificial neural network programs can be used for personification of the therapeutic and diagnostic process and nursing of newborns, including very preterm ones.