Possibilities of neural networks for personalization approaches for prevention of complications after endovascular interventions

Abstract

It is known that most of the diseases of the cardiovascular system are accompanied by disorders in the hemostatic system. The hemostatic system is one of the most complex systems. It has a hierarchical structure with a plurality of components. We analyze the results of thrombin generation test (TGT) which allows of estimating the actions of all components of the hemostatic system. The problem is complicated by the presence of too many various clinical cases. The simple statistical methods do not provide global assessments. We suggest the universal neural network approach for building hemostatic system models based on the factors which don’t have a statistically significant difference for various types of clinical post surgery cases. The neural network instruments allow of taking into account the nonlinear hierarchical nature of considered system and building individual models for each clinical cases. The aim of our study is to develop the neural network hemostatic system model for forecasting of disease progression and complications after endovascular interventions.