多尺度下人体生理时间序列动力特征的提取与分析

发布时间：2017-12-27 08:31

本文关键词：多尺度下人体生理时间序列动力特征的提取与分析　出处：《哈尔滨工业大学》2014年硕士论文　论文类型：学位论文

【摘要】：非线性行为在自然界和社会科学中普遍存在。人体生理时间序列有着十分丰富的非线性行为,具有极其复杂的特性。人体生理时间序列可以很好的反映人体所处的生理状况,及时的发现它们的变化情况,能为人体的一些疾病达到预警的目的。因此,研究人体生理时间序列的动力学行为特征越来越受得医学界和科研人员的关注,并且变得很有现实价值和意义。传统方法往往在单一尺度或者对单一的时间序列进行研究,如:近似熵、样本熵、小波变换模的极大值法、DFA等,不能准确的探测生理时间序列的动力学行为特征,不能较好地反映出人体生理状态。本文采用多尺度熵(M-SE)的方法,在多标度下分析同步采集的不同生理状态下的两列生理时间序列心电信号和呼吸信号的复杂程度。结果表明运动状态下信号复杂度稍微强于平静状态,而健康个体信号复杂度却远远强于心脏疾病个体。采用多重分形去趋势波动分析MF-DFA、多重分形去趋势交叉关联分析MF-DCCA方法,分析不同生理状态下不同时间序列的长期自相关性和交叉关联特性,并且发现心电信号和呼吸信号都具有多重分形的特征。通过不同的参数多重分形谱宽度、多重分形谱面积、多重分形谱对称度的定量计算,描述多重分形属性的强弱,较好的区分人体不同的生理状态。本文在多尺度下运用不同的研究方法探索人体生理时间序列,能够更全面准确的统计分析出人体生理时间序列动力行为特征。对不同生理状态的区分,也取得了较好的成果。文章的理论基础与实际数据分析,也为临床医学中疾病的预防与治疗提供了一个参考方向。
[Abstract]:Nonlinear behavior is common in nature and social science. The human physiological time series has a very rich nonlinear behavior, which has extremely complex characteristics. Human physiological time series can well reflect the physiological condition of human body, detect their changes in time, and achieve the purpose of early warning for some diseases of human body. Therefore, studying the dynamic behavior characteristics of human physiological time series has attracted more and more attention from medical researchers and researchers. The traditional methods are often in a single scale or research on single time series such as approximate entropy, sample entropy, wavelet transform modulus maxima method, DFA, dynamic behavior detection of physiological time series can not be accurate, can not reflect the physiological state of human body. In this paper, we use multi-scale entropy (M-SE) method to analyze the complexity of ECG and respiratory signals in two physiological time series acquired at different physiological levels under multi scale. The results showed that the signal complexity was slightly stronger in the motion state than in the calm state, while the signal complexity of the healthy individuals was much stronger than that of the heart disease. The multifractal detrended fluctuation analysis MF-DFA and multifractal detrended cross correlation analysis MF-DCCA method, analysis of different time series under different physiological conditions of the long-term autocorrelation and cross correlation properties, and found that the characteristics of ECG and respiratory signal has multifractal. Through the quantitative calculation of multifractal spectral width, multifractal spectral area and multifractal spectrum symmetry, we can describe the strength of multifractal attributes and distinguish the different physiological states of human body better. In this paper, we use different research methods to explore human physiological time series under multi-scale, and we can more comprehensively and accurately analyze the dynamic behavior characteristics of human physiological time series. Good results have been achieved in the differentiation of different physiological states. The theoretical basis and actual data analysis of the article also provide a reference for the prevention and treatment of disease in clinical medicine.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R33

【相似文献】