Indices of symbolic dynamic distribution in cardiac patients

Although symbolic dynamics analysis (SDA) has been proposed for encoding words of different length and symbols, the resulting rapidly growing number of patterns has limited its clinical use. Aim of this study is to propose new SDA indices tested on a clinical data-set. We studied 40 ECG Holter of normal (NR), post-MI (MI), heart failure (HF) and transplanted (TR) subjects. RR differences were encoded into 5 symbols, deriving 3, 5 and 7 length words classified by a dominance's criterion in pattern words with a predominant vagal content (V), decelerating content (D), accelerating content (A), sympathetic content (S) and without variability content (0). Their distributions were then quantified by Kurtosis an Chi-square indexes. Results showed an optimum word-length of 3, where both Kurtosis (2.2±0.6; 2.2±0.9; 3.1±0.9; 4.0±0.4) and Chi-square (7±5; 8±5; 26±17; 106±42, respectively for N, MI, HF and TR) showed very significant p<0,0001 values at the ANOVA test among groups, mainly discriminating HF and TR subject by Tukey's post-test. SDA is an helpful technique in interpreting the encoded HRV information. The pattern words distributions clearly tend to lose their tails to the worsening of the autonomic impairment as immediately described by Kurtosy or chi-square index, especially for risk stratification of HF patients.