DOI:https://doi.org/10.15407/kvt199.01.019
Cybernetics and Computer Engineering, 2020, 1(199)
BILOSHYTSKA O.K., Senior Lecturer,
Department of Biomedical Engineering,
e-mail: o.k.biloshytska@gmail.com
NASTENKO Ie.A., DSc (Biology), Senior Researcher
Head of the Biomedical Cybernetics Department
e-mail: nastenko.e@gmail.com
PAVLOV V.A., PhD (Engineering), Associate Professor
Associate Professor of the Biomedical Cybernetics Department
e-mail: pavlov.vladimir264@gmail.com
National Technical University of Ukraine
“Igor Sikorsky Kyiv Polytechnic Institute”,
37, Peremohy av.,03056, Kyiv, Ukraine
THE USE OF COMPLEXITY AND VARIABILITY CHARACTERISTICS FOR THE ANALYSIS OF COMPLEX DYNAMIC SYSTEMS
Introduction. The normal dynamics of a healthy organism is chaotic and the observed “chaos” is inherent in the very nature of the dynamic processes taking place in the organism and the degree of chaotic of these processes may vary in case of pathology in one direction or another. The electrical activity of the brain is also characterized by signs of deterministic chaos, and changes in parameters of its nonlinear dynamics testify to the characteristic changes in brain functioning. The problem of diagnostics and identification of the moment preceding an epileptic seizure or other periods of brain functioning in epileptic patients is not only a problem of choosing a classification method but also of determining quantitative estimates of dynamics reflecting the complexity and variability of the Electroencephalography (EEG) signal.
The purpose of the paper is to form an effective ensemble of features from the characteristics reflecting the complexity and variability of the EEG sig signal ,to construct the prognostic models for the course of epilepsy and to develop the information technology to support diagnostic decision-making based on them.
Methods. The methods of mathematical statistics for the processing of diagnostic information, the methods of mathematical modeling (stepwise logistic regression) — for the construction of prognostic models for estimating the course of epilepsy were used; methodological bases for the creation of information technology for the diagnosis of epilepsy according to the EEG.
Results. Changes in indicators such as Hurst Index, fractal dimension, logistic mapping, and algorithmic signal complexity have been investigated. The mathematical models include variables that are calculated from the EEG data and are available during patient observation. As a result of the application of step-by-step algorithms, the most informative features are included in the models. The selected features allow for the most accurate identification of individual periods of epilepsy flow from the EEG data. It has been established that the use of a decision support system increases the reliability of determining the periods of an epileptic seizure (conditional norm, before, during and after an attack) by an average of 6.6% for children and 8% for adults.
Conclusions. The proposed prognostic models allow to obtain additional information about the periods of epileptic seizures and to predict their onset in time.
Keywords: information technology, EEG, epileptic seizures, epilepsy, complexity and variability indicators, predictive models, logistic regression.
REFERENCES
1 Unified clinical protocol for primary, emergency, secondary (specialized) and tertiary (highly specialized) epilepsy in adults. Kyiv, 2014 (In Ukrainian). URL: http://mtd.dec.gov.ua/images/dodatki/2014_276_Epilepsii/2014_276_YKPMD_epilepsiya_dorosli.pdf. (Last accessed: 11.09.2019)
2 Unified clinical protocol for primary, emergency, secondary (specialized) and tertiary (highly specialized) epilepsy in children. Kyiv, 2014 (In Ukrainian). URL: http://mtd.dec.gov.ua /images/dodatki/2014_276_Epilepsii/2014_276_YKPMD_epilepsiya_dity.pdf. (Last accessed: 11.09.2019)
3 Díaz M.H., Córdova F.M., Cañete L., Palominos F., Cifuentes F., Sánchez C., Herrera M. Order and Chaos in the Brain: Fractal Time Series Analysis of the EEG Activity During a Cognitive Problem Solving Task. 3rd International Conference on Information Technology and Quantitative Management, ITQM 2015. Elsevier. Procedia Computer Science. 2015, Vol. 55, pp. 1410-1419.
https://doi.org/10.1016/j.procs.2015.07.135
4 Wang X., Meng J., Tan G., Zou L. Research on the relation of EEG signal chaos characteristics with high-level intelligence activity of human brain. Nonlinear Biomedical Physics. 2010, no. 4, pp. 2-10.
https://doi.org/10.1186/1753-4631-4-2
5 Mayorov O. Yu. Reliability of bioelectric activity (EEG, ECG and HRV) researches of the deterministic chaos by the nonlinear analysis methods. Conference Paper !·! 3rd Chaotic Modeling and Simulation International Conference (Chania Crete, January, 2010), Chania Crete, Greece, 2010, 61 p.
6 Biloshytska O.K. Investigation of the features of electroencephalographic signals using nonlinear dynamics. Development of science in the XXI century: a collection of articles of the scientific-information center “Knowledge” on the materials of the X international correspondence scientific-practical conference, Part 1 (Kharkiv, 15th of Feb., 2016), Kharkiv, 2016, pp. 11-16. (In Russian).
7 Borisova O.S. Investigation and analysis of native electroencephalographic data by nonlinear dynamics methods: PhD thesis: 05.11.17 / Institute of Technology. Taganrog, 2010, 163 p. (In Russian).
8 Hacken G. Synergy. Moscow, 1980, 404 p. (In Russian).
9 Mayorov O.Yu., Fenchenko V.N. Study of brain bioelectrical activity from the standpoint of multi-dimensional linear and nonlinear EEG analysis. Clinical Informatics and Telemedicine. 2008, no. 4, pp. 12-20. (In Russian).
10 Mayorov O.Yu. Fenchenko V.N. About calculation of parameters of deterministic chaos at research of bioelectrical activity of a brain (EEG). Klin. inform. and telemed. 2006, no. 3, pp. 37-46. (In Russian).
11 Biloshytska O.K., Klymenko T.A. Analysis of the parameters of EEG signals with epileptic activity using nonlinear methods. Bulletin of NTU “KhPI”. Series: Mechanical-technological systems and complexes. 2017, no. 19 (1241), pp. 30-34 (In Ukrainian).
12 Biloshytska O.K. Nonlinear Dynamics as Instruments for Prediction of Pathological Changes in the Electroencephalogram. Bulletin of NTU “KhPI”. Series: Mechanical-technological systems and complexes. 2016, no. 50 (12221), pp.79-83. (In Ukrainian).
13 Biloshytska O.K. Classification of pathological EEG signals using machine learning methods. Bulletin of NTU “KhPI”. Series: Mechanical-technological systems and complexes. 2017, no. 44 (1266), pp.35-39 (In Ukrainian).
14 Biloshytska O.K., Nastenko Ie.A. Evaluation of prognostic possibilities of EEG signal behavior complexity indicators in epileptic seizures. Information systems and technologies in medicine (ISM-2018): Collection of scientific papers of the I International scientific-practical conference (Kharkiv, 28-30th of Nov, 2018), Kharkiv, 2018, pp. 95-97 (In Ukrainian).
15 Kannathal N., Rajendra A., Fadhilah A. Nonlinear analysis of EEG signals at different mental states. BioMedical Engineering OnLine. 2004. URL: http://www.biomedical-engineering-online.com/content/3/1/7. (Last accessed: 02.12.2019 )
16 Gayazova N. T., Zaripov R. P. Stochastic estimation of the rate of human pathological tremor using Hurst index. TGPU Bulletin. 2008, no. 15. (In Russian). URL: https://cyberleninka.ru/article/n/stohasticheskaya-otsenka-skorosti-patologicheskogo-tremora-cheloveka-s-pomoschyu-pokazatelya-hersta (Last accessed: 25.11.2019)
17 Kalush Y., Loginov V. Hurst Indicator and its hidden properties. Siberian Journal of Industrial Mathematics. 2013, no. 4, pp. 29-37. (In Russian).
18 Biloshytska O.K. Use of nonlinear dynamics and cellular-automatic modeling methods to study the dynamic features of the brain neural network. Problems of information technologies. 2015, no.1, pp. 173-180. (In Ukrainian).
19 Borowska M., Oczeretko E., Mazurek A. Application of the Lempel-Ziv complexity measure to the analysis of biosignals and medical images. Annales Academiae Medicae Bialostocensis. 2012, no. 50, pp. 29-30.
20 Garg A., Mathur M., Upadhayay M. Application of LZW Technique for ECG Data Compression. International Journal of Advances in Computer Networks and its Security. 2013, no. 58, pp. 374-377.
21 Strogatz S. Non-linear Dynamics and Chaos: With applications to Physics, Biology, Chemistry and Engineering. Perseus Books, 2000, 505 p.
22 Biloshytska O.K. Information technology for estimating the course of epilepsy by indicators of electroencephalogram complexity: PhD thesis: 05.13.09 / International Research and Training Center for Information Technologies and Systems of the National Academy of Science of Ukraine and Ministry of Education and Science of Ukraine, Kyiv, 2019, 183 p. (In Ukrainian).
23 Gniezditsky V.V. Inverse EEG problem and clinical electroencephalography (mapping and localization of brain electrical activity sources). Moscow, 2004, 624 p. (In Russian).
24 Zenkov L.R. Clinical electroencephalography (with epileptology elements). Manual for doctors. Moscow, 2004, 368 p. (In Russian).
25 Zhirmunskaya E.A. Clinical electroencephalography. Moscow, 1991, 77 p. (In Russian).
26 Biloshytska O.K. Analysis and assessment of non-linear characteristics of epileptic EEG signals. Scientists notes from the VI Tavriya National University Vernadsky. Series: Technical Sciences. 2018, no 29 (68), Part. 1, pp. 80-85. (In Ukrainian).
Received 27.12.2019