VARIABILITY ANALYSIS OF PHYSIOLOGICAL SIGNALS USING NONLINEAR TIME SERIES ANALYSIS TECHNIQUES

Physiological rhythms are not strictly periodic but rather fluctuate irregularly over time. Rhythms interact with each other as well as with the outside fluctuating environment under the control of inestimable feedback systems that provide organized functions that are central to life. Alternations in the rhythms of physiological systems are associated to disease. Physiological systems are complex and science of complex system is closely related to the variability analysis. Variability analysis is the technique of measuring the degree and character of patterns of change of a time series of biological parameter in order to asses the state dynamics of the investigated system.\ud \ud Physiological systems serve as a fascinating playground for the analysis techniques, which stem from the discipline of nonlinear dynamics. The essential nonlinearities and the complexity of physiological interactions limit to the ability of linear analysis to provide full description of the underlying dynamics. This makes nonlinear analysis an invaluable tool for the analysis of physiological signals. Sophisticated and robust time series analysis techniques are needed to quantify the dynamics of physiological signals.\ud \ud This thesis reviews most commonly used measures of variability analysis including time domain, frequency domain and nonlinear measures for heart rate and stride rate variability analysis. This study is methodological approach for quantifying the dynamics of heart rate and stride interval signals in health and disease. The main aim of the study is to develop robust variability analysis measures with improved classification ability. Two nonlinear measures: Threshold based acceleration change index (T ACI) and normalized corrected Shannon entropy (NCSE) at different threshold values have been used to quantify the dynamics of heart and stride interval time series of healthy and diseased subjects. Despite of the fundamental difference in their regulation, the research in heart rate variability analysis has spurred the similar investigations in gait variability analysis.\ud \ud TACI is modified form of acceleration change index (ACI). TACI was proposed to characterize the dynamics of threshold crossing, whereas ACI characterizes the dynamics of zero crossing. In this study the behaviour of TACI for simulated time series (uncorrelated random data and chaotic time series) were studied. Robustness of TACI in the presence of artifacts was investigated. Results are presented for the application of TACI to heart rate and stride interval time series of healthy and diseased subjects.\ud \ud Symbolic time series analysis involves the transformation of the original time series into symbol sequences that can be valuable to extract useful information about the state of the system generating the process. Data symbolization is the first step in symbolic time series analysis. Data symbolization involves the conversion of a data series of many possible values into a symbol series of few distinct values. This coarse graining has practical effect of producing low resolution data from high resolution data. Some detailed information is lost but coarse dynamic behaviour remains and can be analysed. In this study normalized corrected Shannon entropy was used to quantify the dynamics of cardiac interbeat interval and stride interval time series of healthy and diseased subjects at different threshold values. Surrogate data test was performed in order to compare the complexity of physiological signal and random noise