Human Balance Assessment Based On Nonlinear Characteristics Of EEG

The assessment of standing balance ability is the primary task in the treatment of patients with balance disorder,which provides a basis for making reasonable rehabilitation treatment.In this paper,the quantitative analysis of human balance is carried out from the endogenous perspective.This paper discusses the existing methods of balance judgment,those methods completely focus on the external performance of human,and the evaluation results inevitably with partly subjectivity,which leads to the high misdiagnosis rate of human balance assessment and the narrow range of clinical patients.Starting from the central nervous system,this paper explored the brain’s response mechanism of equilibrium,and quantitatively analyzed the characteristics of endogenous EEG signals with different balance abilities from the activation state of cerebral cortex and the characteristics of brain information transmission during the maintenance of equilibrium state.The main work of this paper with the following aspects:Firstly,the balance adjustment physiological mechanism of system has been analysied,and extracted the several important factors influencing human body static state balance,the balance between the sensory inputs on the basis of the design was blocked in experimental paradigm.Which reflect the visual input and proprioception input’s differentiation,what lead to obtain different EEG data.The biggest problem in the experiments is the randomness of the balancing action,that is,it is impossible to predict whether the subjects have made the balancing action at a certain point in the future,so it is necessary to "calibrate" the balancing action of the EEG data.In order to achieve this purpose,the EEG research from the synchronization explained the cerebral cortex in balance adjustment in the process of the actual physical behavior,and defines the concept of balance EEG,then develop an effective balance phase synchronization of brain electrical criteria,accurate quantitative indicators are given,and according to the standards for processing the eeg data collected in the experiments.Thus,a high quality EEG data set was obtained.The results show that the proposed method can greatly avoid the interference of resting EEG and improve the accuracy of evaluation.Based on balanced EEG data,I propose an event-driven transfer entropy network analysis method,and use this method to construct the brain functional network for different experimental paradigms.Compared with the traditional network construction methods,the new method absorbs the unique usage of transfer entropy on the nonestablished mathematical model,and overcomes the problem of entropy accuracy reduction caused by the characteristics of EEG data.In this paper,the traditional twonode causal relationship model is abandoned,the causal relationship between nodes is reduced to correlation,and the study of causal relationship is transferred to the functional brain regions.Calculated by this model avoids the cause and effect in the uncertainty and partial or even contradiction,but from the perspective of functional brain area.The causation can still get good guarantee,embodies the brain regions and multi-source information transfer relationship between brain regions,and based on the synchronization theory put forward by the balance information view have the same starting point,Therefore,it has a good agreement with phase synchronization theory.Finally,it is recognized that the remarkable dynamic characteristics of the brain during balance regulation should contain much balance information,so this is the first time combined with the study of human balance and nonlinear dynamic theory.According to the current research about the dynamics of the brain,puts forward the human balance regulation innovation based on the theory of the black box theory: set up a MISO system of the human body as an initial condition,converting the final balance ability assessment tasks into how to use the system output to determine the initial state of the system.In this paper,lempel-Ziv complexity and Maximun Lyapunov Exponent(MLE)are used as the final classification features,where LZ complexity describes the complexity characteristics of brain under different balance ability states.Reflects the complexity of black box internal system;The most significant feature of MLE index lies in its strong sensitivity to the initial state of the system.MLE can amplify the differences of different levels of balance ability.In conclusion,in this paper,a variety of endogenous features are extracted from EEG data under different equilibrium states from the perspectives of phase synchronization,transfer entropy network and nonlinear dynamics.On this basis,in order to obtain the best endogenous features,combination of multiple classifiers and multiple combination features were used for comparison.The experimental results show that the combination feature [C,E,M] composed of Support Vector Machine(SVM)with network clustering coefficient(C),shortest path(E)and MLE(M)has the best classification performance,which average accuracy of the four classification tasks reaches 76.00%.Finally,this method is used to evaluate the balance of the elderly data and has achieved remarkable results,which proves that the method proposed in this paper is an effective new way to evaluate the balance ability of human.