Symbolic Analysis of Heart Rate Dynamics before the Spontaneous Onset of Paroxysmal Atrial Fibrillation

ABSTRACT

BACKGROUND AND OBJECTIVES: Heart rate variability (HRV) analysis, by conventional measures, for predicting the risk of atrial fibrillation (AF) has inherent shortcomings. Recently, nonlinear HRV analysis methods have been developed to reveal heart rate dynamics not evident from usual HRV measures. The purpose of this study was to test the hypothesis"heart rate dynamics are altered before the spontaneous onset of paroxysmal AF (PAF)"using algorithms derived from symbolic dynamics based on nonlinear system theory. SUBJECTS AND METHODS: This study included 34 subjects (30 males, 59+/-9 years) 17 PAF patients and 17 gender and age-matched controls, who underwent 24-hour Holter ECG. The dynamics of one hour of normal sinus rhythm before the onset of AF were assessed using 4 different symbolization algorithms, and quantified by Shannon entropy (SE) and Renyi entropy (RE). RESULTS: The SE, RE and Error-corrected SE (ECSE) were larger in the PAF than the control patients when the raw R-R data was assessed (Algorithm I). However, when reconstructed time series, expressed as the time difference between 2 successive R-R interval, were analyzed (Algorithm II, III and IV), the entropy of the PAF patients were consistently smaller than those of the controls. Of the 4 different symbolization algorithms, algorithms II and III showed a significant difference in the SE, ECSE and RE between the PAF and the control patients (p<0.05). The difference in the RE at q=4 was most significant (p<0.001). Of the various word patterns seen during analysis using algorithms II and III, a marker of heart rate variation, the frequency of (1,1,1) pattern, was two-fold higher than in the control patients. The frequency of this pattern was 1.5-fold higher in the PAF patients when algorithm IV was used. These results suggest that a few specific word patterns are repetitive in a pathologic condition, such as PAF, and the heart rate dynamics were decreased in PAF. CONCLUSION: HRV analysis by symbolic dynamics may be a useful adjunctive tool for the detection of abnormal heart rate dynamics as a pathophysiological mechanism of PAF.