DETECTING CHAOS IN HRV SIGNALS IN HUMAN CARDIAC TRANSPLANT RECIPIENTS

A statistical approach to chaos identification in time series is prese nted. The method is applied to numerical data generated by chaotic sys tems and to heart rate variability (HRV) signals of normal subjects an d heart transplant recipients. This method compares the short-term pre dictability for a given time series to an ensemble of random data whic h has the same Fourier spectrum as the original time series. The short -term prediction error is computed as a discriminating statistic for p erforming statistical hypothesis testing. The results suggest that HRV signals of the transplant recipients recorded 3 months after the tran splantations show the same signature of chaos as that of the HRV signa ls for normal subjects. (C) 1997 Academic Press.