Background: We hypothesized that mathematical analysis of atrial fibrillary waves may predict recurrence of atrial fibrillation (AF) after cardioversion. Methods: We included 51 patients with symptomatic persistent AF (40 males, 57 ±8 years). All patients were assessed by holter and echocardiography within 3 months before cardioversion. We extracted only atrial fibrillary waves from V1 lead in holter and measured three parameters including approximate entropy (Aprx), Lyapunov-exponent (Lya) and center-frequency (Cent) to quantify entropy, chaotic activity and frequency, respectively. Recurrence was defined as AF episodes documented by EKG/holter within 30days after cardioversion. Results: The patients with recurrence (n=25) had higher Lya than those without recurrence (n=22)(0.258 ±0.157vs0.161 ±0.148, p=0.036), whereas Aprx was not related to recurrence (p>0.05). The patients with Lya>0.05 had more recurrence than those with Lya<0.05 (64%vs18%, p=0.008), and the patients with Cent<7.0 had more recurrence than those with high Cent>7.0 (68%vs35%, p=0.021). All patients were divided into three groups; groupA, the patients with high-Lya and low-Cent, groupB, those with high-Lya or low-Cent, groupC, those with low-Lya and high-Cent. GroupA showed more recurrence than groupB, and groupB did more recurrence than groupC. By logistic regression, Lya and Cent scale were strong independent predictors of recurrence (OR=6.9, CI:95% 2.0~23.5, p=0.002). Conclusions: Chaotic activity with low frequency in fibrillary waves could predict AF recurrence within 1month after cardioversion.
- atrial fibrillation
- lyapunov exponent