Прогнозирование инифицированного панкреонекроза методом нейронных сетей

Abstract

The purpose of the research: 1. to create a scale to predict infection of pancreatic necrosis on the early stages of hospitalization; 2. to create a neural network on the basis of the constructed scale. We used statistical methods: correlation analysis, regression analysis, discriminant analysis. The result of the research is a new scale, which allows to predict infection of pancreatic necrosis on the early stages of hospitalization. This scale includes 6 indicators calculated on the complete blood count of patients at entering in-patient department and in the first twenty-four hours. On the basis of the constructed scale the neural network was created, it allows to produce more accurate prediction with the help of computer technologies. The main advantages of the proposed methods are a high exactness of prognosis, possibility of the use since the first days of disease and simplicity of definition of necessary indicators