摘要
为了准确分辨并识别不同人体的步态特征,提出采用递归图和递归定量分析的方法,计算人体步态非线性时间序列的复杂度。首先利用互信息和伪邻近法分别计算得到合适的延时时间和嵌入维数,根据相空间重构的原理将一维时间序列扩展到高维相空间中,获得时间序列在高维空间中邻近点的分布规律和运动特点。构建了患有帕金森疾病的老年人、健康老年人和健康年轻人的步态信号递归图,可以直观定性分析和评估这3组人群的步态信号的空间分布程度,其中健康人群最复杂。采用递归定量分析,量化了人体步态的复杂度,结果表明,患有帕金森疾病的人群的步态复杂度最小,而且独立样本t检验显示了3组人群的复杂度具有显著的差异性。该方法简单可行,可以准确地对不同年龄和帕金森疾病的人群进行分类识别,有利于人体健康监测和诊断研究。
To accurately classify and identify different human gait signals,recurrence plot and recurrence quantification analysis are proposed to calculate the complexity of nonlinear time series underlying human gait stride interval.The appropriate delay time and embedding dimension are respectively calculated by methods of the mutual information and the false nearest neighbours.One dimension time series is extended to high dimension phase space by employing phase space reconstruction principle for studying the distribution regularity and movement characteristic of neighboring points in time series.The recurrence plots of gait signals from Parkinson＇s disease elders,healthy elders and youth are established for intuitively and qualitatively analyzing and evaluating their distribution state.It is observed that the space distribution of healthy subjects is more complex than others.Moreover,the complexity of human gait is quantified by recurrence quantification analysis.The computation results show that gait complexity of subjects with Parkinson＇s disease is less than healthy youth and elders,and independent sample t-test also exhibits their significant difference for distinct classification.Consequently,the proposed method is easy and feasible for classifying the subjects with different ages and Parkinson＇s disease.It isalso helpful to human health monitoring and diagnosis.
作者
余建
曹军义
王伟
廖维新
YU Jian;CAOJunyi;WANG Wei;LIAO Wei-Hsin(State Key Laboratory for Manufacturing Systems Engineering, Xi（n Jiaotong University, Xi’ an 710049, China;Department of Mechanical and Automation Engineering, The Chinese University of Hong Kong, 999077,China)
出处
《西安交通大学学报》
EI
CAS
CSCD
北大核心
2017年第10期47-52,70,共7页
Journal of Xi'an Jiaotong University
基金
国家自然科学基金资助项目(51575426,51611530547)
中央高校基本科研业务费综合交叉项目(xjj2016002)
关键词
递归图
递归定量分析
人体步态
复杂度
健康监测
recurrence plot
recurrence quantification analysis
human gait
complexity
healthmonitoring
