One problem in atrial fibrillation detection is that respiratory cycles induce a variation in the inter-beat interval (IBI) signal (140) determined from a cardiac activity signal (130). This results in a large number of false detected episodes. Analyzing the variability and/or chaotic structure of an IBI signal (140) as well as the frequency pattern of said IBI signal (140), i.e., the respiratory frequency power (180), e.g., in the power spectral density (PSD) spectrum (160) of said IBI signal (140), allows reducing the number of false detected episodes of atrial fibrillation.
