Automatic detection of atrial fibrillation using R-R interval signal

This paper presents a method to automatically detect atrial fibrillation (AF) based on the scatter plot of R-R interval signal. R-R intervals of AF are absolutely irregular, which are different from those of normal ECG. Scatter plot of R-R interval signal is illustrated here and characteristic indexes are extracted from the plot, i.e. VAI, VLI, SD1, and SD2. Algorithm test using AF Termination Challenge Database and clinical ECG data shows that these four indexes have high sensitivity and specificity to classify AF from normal ECG. Then a conjoint analysis method to combine these four indexes is proposed to detect AF with more high sensitivity and specificity.