人体静态平衡运动力学分析中多元多尺度熵特征改进算法

Improved Multivariate Multi-scale Entropy Feature Algorithm for Mechanic Analysis of Static Balance of Human Body

目的利用改进的多元多尺度熵特征对人体静态平衡运动力学信号进行分析。方法在多元多尺度熵计算中,要对多元延迟向量的维数进行扩展,本文针对性地将单个变量的依次嵌入改成对所有变量同时嵌入。结果改进算法的多元多尺度熵特征应用于多种平衡模式的实验中,处理速度更快,熵值在模式间的距离更大,模式内的散度更小,更易区分。结论本文算法提高了计算效率,改善了特征的可区分性,可以更好地分析人体的静态平衡能力。

Objective To analyze the mechanic signal of static balance of human body by using improved multi- variate multi-scale entropy （MMSE）. Methods When the dimension of MMSE increases, the multivariate de- lay vector needs to be extended. Improvement for the traditional method was proposed in this paper where all vectors were embedded at the same time instead of embedded one by one in the traditional algorithm. Results The experimental results of MMSE feature in different modes showed that the processing speed of improved al- gorithm was faster, the separation distance of entropy value in the different modes was bigger, the dispersed degree was lower, and the feature was easier to distinguish than that of the traditional algorithm. Conclusion The improved algorithm proposed in this paper could elevate the computational efficiency and make the distin- guishability of the feature better, and can better analyze the static balance ability of the human body.