The OMIS Software System: A Module for Analysis of Physiological Processes

Abstract

Analysis of the time series of RR-intervals in electrocardiograms (ECG) is the main method of studying the dynamics of the sinus rhythm [2, 14, 15, 19, 21, 27]. Dynamics of RR-intervals is described using standard statistical parameters (simple mean, variance, etc.) and spectra within various frequency ranges (LF, HF, VLF) [2, 7, 15, 19, 27]. Recently, several techniques for automated analysis of the dynamics of other ECG elements have been suggested [24, 26, 29, 30]. Methods for computer analysis of electroencephalograms (EEG) are usually limited to analysis of the frequency spectrum within the ranges of α-, β-, γ-, δ-, and θ-waves and analysis of correlation (cross-correlation functions) between instantaneous values of the source processes [3, 17]. In some cases, the distribution of intervals between the points of EEG intersection with the zero level [4], the amplitude distribution, and some other distributions are analyzed. However, these methods of analysis do not take into account a fundamental property of physiological processes: the relationship of various phases of individual activity cycles to the functional state of the human body [9, 11, 25]. Only methods taking into account the qualitative nonuniformity of the dynamics of physiological processes (inhalationnexhalation, systolendiastole, muscular contraction and relaxation, etc.) can provide comprehensive analysis of these processes. To study the dynamics of various physiological processes it is necessary to know how the preceding phase of a process affects the following phase. Nonlinear dynamics methods are insufficient to solve this problem. These methods use an autocorrelation function and its spectrum for assessing the fractal dimensionality, so that the information about the fine phase structure of the oscillation process is lost [20]. Methods based on clinical knowledge (typological classification of EEG [13] or ECG [22, 23] or structural description of the ECG elements and the cardiosignal as a whole [5]) also do not provide adequate description of the temporal dynamics of physiological processes. The methods of analysis of the temporal dynamics of physiological processes considered in this work are based on dividing the source process into individual oscillations and generating discrete series of various characteristics of these individual oscillations. The resulting time series are considered as elements of the source physiological process as a systemic whole [9, 11, 25].