DOI:https://doi.org/10.15407/kvt198.04.003
Cybernetics and Computer Engineering, 2019, 4(198)
PEROVA I.G., PhD (Engineering), Associate Professor
Associate Professor of Biomedical Engineering Department,
e-mail: rikywenok@gmail.com
BODYANSKIY Ye.V., DSc (Engineering), Professor
Professor of Artificial Intelligence Department
e-mail: yevgeniy.bodyanskiy@nure.ua
Kharkiv National University of Radio Electronics,
14, Nauky av., Kharkiv, 61166, Ukraine.
ONLINE MEDICAL DATA STREAM MINING BASED ON ADAPTIVE NEURO-FUZZY APPROACHES
Introduction. Data mining approaches in medical diagnostics tasks have a number of special properties that do not allow the use of such approaches in a classical form. That’s why adaptive neuro-fuzzy systems for online medical data stream processing tasks and its learning algorithms have been developed. Proposed systems can process medical data streams in three modes: supervised learning, unsupervised learning and active learning.
The purpose of the paper is to develop approach, based on adaptive neuro-fuzzy systems to solve the tasks of medical data stream mining in online-mode.
Methods. The methods of computational intelligence are used for medical data stream processing and, first of all, artificial neural networks, neuro-fuzzy systems, neo-fuzzy systems, their supervised, unsupervised and active learning approaches, gradient methods of optimization, methods of evolving system.
Results. As a result, approbation of the developed approach in supervised learning mode using multidimensional neo-fuzzy neuron on medical data of patients with urological disease was investigated. Percentage of errors in system testing using all feature space is 11.11 %, using the most informative features the error rate becomes 6.4 %. Also multidimensional neo-fuzzy neuron was used for diagnostic of the pharmacoresistant form of epilepsy, percentage of errors in system testing is 5.82 %. Approval of the developed approach in the mode of active training and association on the data of patients with pulmonary diseases was performed. For all approbation results performance criterion was calculated, its values are suitable for the tasks of medical diagnostics in data stream mode.
Conclusions. The proposed neuro-fuzzy approaches allow obtaining additional information about patients’ diagnosis in conditions of limited a priori information about patient.
Keywords: adaptive system, neuro-fuzzy system, medical data mining, medical data stream.
REFERENCES
1. de Oliveira J., Pedrycz W. Advances in Fuzzy Clustering and its Applications. 2007.
2. Berka P., Rauch J., Zighed D. Data mining and medical knowledge management cases and applications. New-York, 2009. https://doi.org/10.4018/978-1-60566-218-3
3. Giannopoulou E. Data mining in medical and biological research. New York, 2008.
4. Karahoca A. Data Mining Applications in Engineering and Medicine. InTechOpen. 2012. https://doi.org/10.5772/2616
5. Han J.,Kamber M. Data Mining: Concepts and Techniques. Amsterdam, 2006.
6. Gorban A., Kegl B., Wunsch B., Zinovyev A. Principal Manifolds for Data Visualization and Dimension Reduction. Lecture Notes in Computational Science and Engineering. Berlin-Heidelberg-New York, 2007.Vol. 58. https://doi.org/10.1007/978-3-540-73750-6
7. Mulesa P., Perova I. Fuzzy Spacial Extrapolation Method Using Manhattan Metrics for Tasks of Medical Data Mining. Proceeding of Computer Science and Information Technologies CSIT’2015 (Lviv, 14-17th of Sept, 2015). Lviv, 2015. P. 104-106. https://doi.org/10.1109/STC-CSIT.2015.7325443
8. Rastrigin L.A. Adaptation of complex systems. Riga,1981. (in Russian).
9. Bodyanskiy E., Rudenko O. Artificial neural networks: architectures, training, applications. Kharkiv, 2004. (in Russian).
10. Fainzilberg L. Mathematical methods for evaluating the utility of diagnostic features. Kyiv, 2010. (in Russian).
11. PerovaI., BodyanskiyY. Adaptive human machine interaction approach for feature selection-extraction task in medical data mining. International Journal of Computing. 2018. Vol. 17. No 2. P. 113-119.
12. Oja E., A simplified neuron model as a principal component analyzer. J. of Math. Biology. 1982.No 15. P. 267-273. https://doi.org/10.1007/BF00275687
13. Oja E. Neural Network, principal components and subspaces. Int. J. of Neural Systems. 1989. P. 61-68. https://doi.org/10.1142/S0129065789000475
14. Oja E. Principal component, minor components, and linear neural networks. Neural Networks. 1992. no. 5. P. 927-935. https://doi.org/10.1016/S0893-6080(05)80089-9
15. Bodyanskiy Ye., Perova I., Zhernova P. Online fuzzy clustering of high – dimensional data based on ensembles in data stream mining tasks. The Current State of Research and Technology in Industry. 2019. No1(7). P. 16-24. https://doi.org/10.30837/2522-9818.2019.7.016
16. Yamakawa T., Uchino E., Miki T. Kusanagi H. A neo fuzzy neuron and its applications to system identification and prediction of the system behavior. Proceeding 2nd Int. Conf. on Fuzzy Logic and Neural Networks. July 1992. Iizuka, Japan. 1992. P. 477-483.
17. Landim R., Rodrigues B., Silva S., Matos W. A neo-fuzzy-neuron with real- time training applied to flux observer for an induction motor. Proceeding V-th Brazilian Symp. on Neural Networks (Los Alamitos, CA 04th – 06th of Nov, 1998). Los Alamitos, 1998. P.67-72.
18. Mahmoud S., Perova I., Pliss I. Multidimensional neo-fuzzy-neuron for solving medical diagnostics tasks in online-mode. Journal of Applied Computer Science. 2017. Vol. 25. No 1. P. 39-48.
19. Perova I., Pliss I. Deep hybrid system of computational intelligence with architecture adaptation for medical fuzzy diagnostics. I.J. Intelligent System and Applications. 2017. Vol. 7. P. 12-21. https://doi.org/10.5815/ijisa.2017.07.02
20. Rizzo R. Computational Intelligence Methods for Bioinformatics and Biostatistics. . In Lecture Notes in Bioinformatics: 7th International Meeting, CIBIB 2010. (Palermo, Italy, 16-18th of Sept, 2010). Palermo, 2010. 2011. 301 p.
21. Kountchev R. Advances in Intelligent Analysis of Medical Data and Decision Support Systems (Studies in Computational Intelligence). Berlin Heidelberg, 2013. 246 p. https://doi.org/10.1007/978-3-319-00029-9
22. Cichocki A.,UnbehauenR. Neural Networks for Optimization and Signal Processing. Stuttgart, 1993. 526p.
23. Perova I.,BodyanskiyY., Adaptive fuzzy clustering based on Manhattan metrics in medical and biological applications. Newsletter of the National University “Lviv Polytechnic”. Vol. 826. 2015. P. 8-12.
24. Lughofer E. Evolving Fuzzy Systems – Methodologies, Advanced Concept and Applications. Berlin-Heidelberg, 2011. 454p. https://doi.org/10.1007/978-3-642-18087-3
25. Lughofer E. Single pass active learning with conflict and ignorance. Evolving Systems. 2012. Vol.3. No4. P. 251-271. https://doi.org/10.1007/s12530-012-9060-7
26. Pima Indians Diabetes dataset. Available from URL: http://archive.ics.uci.edu/ml/machine-learning-databases/pima-indians-diabetes/pima-indians-diabetes.data. (Last accessed: 01.05.2008.)
27. Dermatology dataset. URL: http://archive.ics.uci.edu/ml/machine-learning-databases/dermatology/dermatology.data. (Last accessed:01.05.2008.)
28. Parkinson dataset. URL: http://archive.ics.uci.edu/ml/machine-learning-databases/parkinsons/parkinsons.data. (Last accessed:01.05.2008.)
29. Iris dataset. URL: http://archive.ics.uci.edu/ml/machine-learning-databases/iris/iris.data. (Last accessed: 01.05.2008.)
Received 10.07.2019