Advanced spectral methods for detecting dynamic behaviour

The traditional analysis in the frequency domain of cardiovascular variability signals requires stationarity along the considered temporal window, in order to obtain reliable indicators of the sympatho-vagal balance (low frequency (LF) and high frequency (HF) power and frequency, and LF/HF ratio). Through proper advanced algorithms of signal processing, it is possible to implement methods that allow the enhancement of important parameters about the behaviour of the system under investigation in the time and frequency domain. Both non-parametric and parametric time–frequency methods are generally employed at this purpose. Among them, Wigner–Ville Distribution and Time-Variant Autoregressive models are here described. Through such advanced methods of signal processing, it is possible to investigate the dynamic properties of the spectral parameters during transient physiological or pathological episodes, after a proper validation using simulated signals. The methods are used in various applicative areas of interest where the spectral parameters present a significant change in time and where the classical spectral analysis cannot be correctly applied. A few significant cases will be discussed such as tilting manoeuvre, vaso-vagal syncope onset and progression, and acute ischemic episodes. Further, multivariate analysis can be applied in which the focus is on squared coherence function and phase relationships, in order to estimate some possible causal effects in different experimental conditions. It is believed that such advanced methods of time-variant or time–frequency approaches are capable of overcoming the problem of stationarity in classical spectral analysis and to make applicable frequency domain techniques in the study of transient episodes which generally characterise various physiological and clinical conditions.