Acta Biomedica Scientifica (May 2013)
Use of artificial neural networks for differentiated diagnostics of ischemic and hemorrhagic perinatal affections of central neural system in newborns of different terms of gestation
Abstract
The article describes neonatal and pediatric neurology researching. Among the causes of childhood disability first place belongs to diseases of the nervous system. Among perinatal brain damage leading place is occupied by cerebrovascular pathology. One of the main causes of hemorrhagic and ischemic brain damage is impaired cerebral hemodynamics. However there is no single point of view on the processes underlying the development of ischemic brain lesions and intracranial hemorrhage in premature infants. It reveals necessity of immunobiochemical neurospecific proteins defining during neonatal period. Proteins, namely neurospecific enolase, a neurotrophicfactor of nerve growth, vascularendothelial growth factor, allow early finding of pathological disorder. What is a profitable advantage compared to the widely used clinical and instrumental examination and laboratory methods to assist in determining location and extent of the brain. Articleshows importance for a multifunction-oriented model of studying peculiarities of the child, starting with finding patterns in complex processes, due to the influence of internal and external factors on the functional state of the organism based on its individual characteristics, and ending with the solution of problems of differential diagnosis. Thus enabling to seek for hidden dependencies in complex processes conditioned by internal and external factors, leading us to performing differential diagnosis. As for mathematical models and data processing algorithms, the authors used an artificial neural network. These algorithms are used when there is no a precise decision-makingsystem. The medical diagnosis of ischemic and hemorrhagic perinatal central nervous system lesions of newborns maybe added in the list problems to be solved by artificial neural networks. The paper gives valuable information aboutinvestigating child's body properties with neural networks algorithms. Results of applying these algorithms are aimed to increase accuracy of differential diagnosis of ischemic or hemorrhagic perinatal damage to the central nervous system in newborns of different gestational ages are presented.
