Klasifikace komplexních metabolických pacientských dat

Abstract

This bachelor thesis deals with software development in MATLAB programming environment, which is able to classify a large dataset of complex metabolic pacient’s data. The dataset was obtained in clinical examination of the pacients in Hradec Kralove hospital in years 2007-2010. Programme is based on the methods of artificial inteligence and its core is formed from the self-organizing map neural network architecture. Application is designed to process multi-dimensional input data and classify them into classes with similar parameters, so-called clusters. From graphical and numerical programme‘s outputs are doctors able to determine, which metabolic parameters are highly dependent and which parameters have no mutual relations. Afterwards it is possible to create typology of the pacients and evolve more effective therapy. The whole pacient’s dataset is stored in MySQL database system. Design and formation of the database is also part of this bachelor thesis.