Shatnyi S. Information technology cardio signals processing and analysis using a neural network

The dissertation is devoted to development and improvement of models, methods and means of information technology of electrocardiogram processing, increase of speed and accuracy of processing of cardio signals, reduction of the size of the system intended for such processing, reduction of its power consumption and realization of system in analog and digital element bases. The relevance of the topic of the dissertation is substantiated, the purpose and main tasks of research are formulated, the scientific novelty of the work and the practical significance of the obtained results are determined, the connection of the work with scientific topics is shown. Information on approbation of the results of the work, personal contribution of the author and his publication is given. It was found that the efficiency of processing and analysis depends on the quality of signal pre-processing and the nature of the signal itself. Analysis of approaches to the construction of biomedical signal processing systems has shown the need to increase their efficiency. The results of the analysis of the existing cardio signal processing systems allowed us to state that in most of them the classification accuracy is not high enough (not higher than 75%), low speed and high cost of equipment due to the monopoly of the manufacturing companies. The developed method of electrocardiogram analysis by determining the amplitude and duration of each of the P, Q, R, S, T-segments is presented. The method of pre-processing of cardio signals due to its use for identification and filtering of neural networks has been improved. The method of classification of cardio signals by means of use of a partially-parallel neural network is improved. Software and hardware implementations of information technology of cardio signal processing, structural and functional schemes of input signal processing on the basis of microcontrollers and programmable logic integrated circuits are developed. Modeling and optimization of data exchange between the structural elements of the system. Developed specialized software products for ECG pre-processing and analysis. Server tools for the operation of a remote web-system for the interaction of the logical model "doctor-patient" have been developed.