Session P81.2

Detrended Fluctuation Analysis of Heart Rate by Means of Symbol Series

JF Valencia*, M Vallverdú, R Schroeder, A Voss, I Cygankiewicz,
R Vázquez, A Bayés de Luna, P Caminal

Universitat Politecnica de Catalunya
Barcelona, Spain

Detrended fluctuation analysis (DFA) has been shown to be a useful tool for diagnosis in patients with cardiac diseases. The scaling exponents obtained with DFA are an indicator of power-law correlations in signal fluctuation, independent of signal amplitude and external trends. In this work, an approach based on DFA was proposed for analyzing heart rate variability (HRV) by means of RR series which are characterized by long-range correlations. The approach consisted of transforming consecutive RR differences to symbols according to an adapted symbolic-quantization based on mean and standard deviation of the series. DFA was applied to heart beat series as: a) RR series, b) RR increment series, c) magnitude of RR increment series, d) sign of RR increment series and e) symbol series obtained from cases (b) and (c). Three scaling exponents were calculated, aHF, aLF and aVLF, which correspond to the well known VLF, LF and HF bands in the power spectral of HRV. Traditional parameters of time and frequency domain suggested by the HRV Task Force were also calculated. 24-hour RR series of 222 age-matched patients with ischemic cardiomyopathy were analyzed. The end-point was patients that suffered cardiac mortality (CM) after a follow-up of three years. The study considered 30 patients with CM as a high risk group and 192 survivor patients (SV) as a low risk group. Statistically significant differences with p<0.05 were taken into account for statistical analysis of risk groups. The results showed a high correlation between the scaling exponents and the traditional frequency domain parameters. Scaling exponent values in SV group were higher than in CM group for aHF and aLF, and aVLF showed an apposite behaviour. These indicated a reduction of the long-range correlation in high risk group when it was compared with low risk group, in HF and LF bands. Our proposal allowed a better statistical differentiation between both groups of patients, SV and CM (aHF: p=0.0027; aLF: p=0.0009; aVLF: p=0.0141). Parameters of time and frequency domain were not able to statistically differentiate both groups. In conclusion, it seems that an adequate symbolic transformation of the RR series allows DFA scaling exponents to better characterize different correlation properties between the studied cardiac risk groups.

(Abstract Control Number: 147)