Nonlinear Dynamic Analysis Based On Gait Signals Intelligent Classification Of Neurodegenerative Diseases

With the development of the aging society,the population of neurodegenerative diseases is increasing.However,the diagnosis and evaluation of gait disorders in neurodegenerative diseases are mainly based on clinicians' visual observation or drug testing methods.The above methods are easily influenced by the experience of clinicians,so they are somewhat subjective and lack objective standards.In recent years,the quantitative analysis methods based on sensor technology has gradually become a hot spot.This study applies the emerging nonlinear dynamics theory to the analysis of gait signals of neurodegenerative diseases collected by sensors,which can reflect the intrinsic properties of gait signals,thus effectively detecting and analyzing neurodegenerative diseases and establishing a relatively objective standard to assist in the diagnosis of gait signals in neurodegenerative diseases.Based on previous studies,the main results of this article are as follows:(1)The inherent nonlinear dynamic characteristics of time-varying gait time series are excavated.The signal researched is a gait time series,which is obtained by the ground reaction force measured by the pressure sensitive insole in the shoe,and the left foot stride time signal curve,the right foot stride time signal curve,the left foot standing time signal curve,the right foot standing time signal curve and the two foot support time signal curve are extracted from it.Based on the theory of nonlinear dynamics,the nonlinear dynamic characteristics of time-varying gait time series are analyzed.Firstly,by using the power spectrum,the peak of the power spectrum of gait time series is connected into one piece and continuous,and it is preliminarily judged that the signal has chaotic characteristics.Secondly,the principal component spectrum obtained by the principal component analysis method contains a straight line part,and it is determined that the collected time-varying gait signal has nonlinear chaotic characteristics.At the same time,the differences of nonlinear dynamic characteristics between normal individuals and patients with three typical neurodegenerative diseases were observed.(2)Gait feature extraction based on nonlinear dynamics.Firstly,LZ complexity and C0 complexity are extracted from the point of view that complexity is the disorder degree of time series.Secondly,approximate entropy,fuzzy entropy,sampleentropy and wavelet entropy are extracted from the perspective that entropy is a measure of the probability of generating a new pattern in a time series.Finally,the wavelet coefficient features are extracted from the gait signal from the point of view of time frequency,and all the values of each feature are statistically analyzed.By analyzing the distribution of features between each type of patients with neurodegenerative diseases and the control group,the validity of each feature for the classification of neurodegenerative disease detection is preliminarily analyzed.(3)Detection of neurodegenerative diseases based on nonlinear kinetic characteristics.First,this study compares the single features under different classifiers,compares the classification results of each feature,and simultaneously compares the impact of different classifiers on the classification results.After fusion of multiple features,the classification analysis is carried out,and it is found that the classification sensitive features between 7 different types(NDDs and CO,ALS and CO,PD and CO,HD and CO,ALS and PD,ALS and HD,PD and HD)are not the same.Through the forward feature selection algorithm and the m RMR algorithm,the features with small correlation in each type are filtered,which improves the correct rate of each classification type.Finally,the database of neurodegenerative diseases is classified in many categories based on support vector machine.(4)Based on MATLAB software and MySQL database,a visual multi-classification system for neurodegenerative diseases database was developed.This system realizes the import of the gait database and the original gait signal graphics display,carries on the signal preprocessing and displays related graphics,then realizes the feature extraction and the feature selection,finally achieves the data management,the neurodegenerative disease automatic detection function.This system is practical and easy to operate,and it can provide doctors with quantitative indicators for reference to diagnose neurodegenerative diseases,so as to better serve medical staff.