Оптимизация метода родоразрешения женщин при развитии аномалий родовой деятельности

Abstract

The aim of the research. To develop a means that would optimise the method of delivery for women with development of abnormalities of labour. Material and methods. Retrospective analysis of 186 cases of labour was carried out at the premises of the perinatal centre of the Regional Clinical Hospital in Chita in 2018-2021. The total sample was divided into 5 study groups: group 1 included 33 women with primary uterine inertia; group 2 included 16 women with secondary uterine inertia, group 3 included 32 women with discoordinated labour activity, group 4 included 55 women with excessive uterine activity, group 5 included 50 women with abnormal uterine contractile activity which arose against the background of the clinical narrow pelvis. All women on the eve of labour (1-2 days) underwent general and special obstetric examination. The groups were comparable in nationality, age, financial and social conditions of the women, their gestational age, frequency of genital and extragenital pathology as well as the main pathology of pregnancy. Statistical processing of the results was carried out using the IBM SPSS Statistics Version 25.0 program. Results. Optimisation of the delivery method for patients with abnormal labour was implemented on the basis of multilayer perceptron, the percentage of incorrect predictions in the learning process of which was 2.4%. The structure of the trained neural network included 14 input neurons: gestational age, labour parity, the woman’s height, uterine fundus height, presence / absence of fetal growth retardation or macrosomia, oligohydramnios, prelabour rupture of amniotic membranes and the posterior view of the foetus’s occipital presentation. Conclusion. An integrated approach based on implementation of universally available methods of objective and instrumental research on the eve of labour makes it possible to choose the correct method of delivery for patients with abnormalities of labour with an accuracy of 96.2%.