Working on the Noltisalis database: measurement of nonlinear properties in heart rate variability signals

Abstract

We present results obtained from the analysis of 50 heart rate variability series (HRV) which have been extracted from Holter recordings in the 24-hours in normal subjects and pathological patients. Data have been collected inside a multicentric research program, which aimed at the nonlinear analysis of HRV series. Multifractal approaches such as generalized structure functions have been used to characterize the HRV signal. Moreover, classical parameters for the analysis of the HRV signal over long time scales have been considered to perform a proper comparison. We considered classical time-domain indexes, "monofractal" characteristics (1/f/sup /spl alpha// spectrum; detrended fluctuation analysis) and a regularity statistic (approximate entropy). The hypothesis of nonlinearity for the HRV signal has been verified by computing the generalized structure function on a set of surrogate data (amplitude adjusted surrogate data). In most cases, the multifractal spectrum of the original HRV series significantly differs (t-test), from those obtained from surrogate signals. This result can be associated with the presence of nonlinear correlations in the HRV signal. Moreover, results show that nonlinear parameters can be used to separate normal subjects from patients suffering from cardiovascular diseases.