Session S22.4

Rhythmometric Analysis of Heart Rate Variability Indices during Long Term Monitoring

R Goya-Esteban*, I Mora-Jiménez, JL Rojo-Álvarez, O Barquero-Pérez, S Manzano-Martínez,
F Pastor-Pérez, D Pascual-Figal, A García-Alberola

University Rey Juan Carlos
Fuenlabrada, Spain

Recently, long term monitoring of ECG signals is receiving much attention. How to deal with this massive amount of information is still an open issue. In particular, Heart Rate Variability (HRV) markers have been widely used to characterize the state of the autonomous regulation of the heart from 24 hours Holter monitoring, but there is few knowledge on the long-term evolution of HRV indices. A data set of 7-days Holter recordings in 14 Congestive Heart Failure (CHF) patients was assembled. For its analysis, an automatic rhythmometry procedure was designed, allowing the characterization of the ultradian and the infradian components, with possible inclusion of near-periodic fluctuations. A bootstrap hypothesis test allows us to systematically adjust the model architecture for each patient. The temporal evolution of relevant time-domain (AVNN, SDNN, NN50), frequency-domain (LF, HF, LFn, LF/HF), and nonlinear (alpha1, SampEn) HRV indices, was analyzed with this technique in CHF recordings. Percentage of power explained by the model was larger in AVNN (47.4+-22.6% for patients), NN50 (40.6+-16.6%), HF (36.2+-14.2%), and SampEn (34.3+-16.7%). Near-periodic fluctuations were mostly present in NN50 (50% patients), SampEn (28.6%), HF (21.4%), and alpha1 (21.4%). Subharmonics were mostly present in NN50 (42.9% patients), AVNN (28.6%), alpha1 (28.6%), and SampEn (28.6%). Larger relative deviations from the daily average pattern were more clearly observed in nonlinear indices and in NN50. Infradian subharmonic was markedly present in LFn, NN50, alpha1, and SampEn, but not in the other indices. We conclude that long term monitoring of HRV can be readily addressed with the automatic rhythmometry procedure.

(Abstract Control Number: 146)