CASPIAN JOURNAL

MANAGEMENT AND HIGH TECHNOLOGIES

NEURAL NETWORK ANALYSIS OF INTERACTION OF POSTURAL CONTROL MACHANISMS UNDER ARTUFICIAL FEEDBACK

Read Gorshkov Oleg G., Starchenko Irina B., Sliva Andrey S. NEURAL NETWORK ANALYSIS OF INTERACTION OF POSTURAL CONTROL MACHANISMS UNDER ARTUFICIAL FEEDBACK // Caspian journal : management and high technologies.  2016.  2.  pp. 25-36.

Gorshkov Oleg G. - Teacher, Donetsk National Medical University, 16 Ilich ave., Donetsk, 83003, Donetsk People's Republic, olgor22@yahoo.com

Starchenko Irina B. - D.Sc. (Engineering), Professor, Southern Federal University, 2 Shevchenko st., Taganrog, 347922, Russian Federation, star@fep.tti.sfedu.ru

Sliva Andrey S. - post-graduate student, Southern Federal University, 2 Shevchenko st., Taganrog, 347922, Russian Federation, stabilan@orbritm.com.ru

In the study of the system of movement’s regulation a key role belongs to the study of the maintenance of the vertical posture. J.J. Collins and C.J. De Luca made significant contribution to the study of the mechanisms of human upright posture regulation. They offered a technique of an estimation of fractal properties stabilograms - stabilogram diffusion analysis (SDA), which allowed revealing two of postural control mechanism "open loop" and "closed loop". In this paper we studied the interaction of postural control (PC) mechanisms "open loop" and "closed loop" in anteroposterior and mediolateral directions under artificial strengthening of the PC due to the introduction of artificial feedback. To identify the statistical significance of the PC mechanisms, which improve postural balance (PB), the logistic regression model of prediction with the integrated assessment of postural movement (PM) was constructed. In order to determine the interaction of PC mechanisms between them under condition of artificial improvement of PB, neural network model, estimating PD in versions with artificial feedback and without it, was constructed. As a result, the obtained neural network model describes the interaction of the components that improve PB due to the introduction of artificial feedback. The sensitivity of the model on the training set was 88,2 % (95 % confidence interval (CI) for interval 80,5-94,2 %), specificity - 88,9 % (95,% CI for interval 81,0-94,8 %). The result can be used to create the devices (hardware) to increase BP in rehabilitation purposes.

Key words: , , , , , , , , , , SDA , fractal analys