Automatic Filtering of R‐R Intervals for Heart Rate Variability Analysis

Background: In R‐R analysis for heart rate variability, rhythm modificatipns other than those related to autonomic nervous system activity must be avoided. We describe an algorithm that performs a correction based on the comparison of each interval with a normal standard. Methods: Starting with a given R‐R list, a parameter R between 0 and 1 is chosen, the normal R‐R range is defined as: (1‐R, 1+R]. The first interval then is divided by the mode of the entire R‐R list. If the resulting value lies within the normal range, then it could be accepted as normal. Subsequent intervals are divided by the last accepted one and the mean value of the accepted intervals. If either of these results is inside.the normal range, then the considered interval is accepted. The abnormal intervals are evaluated to find out if they are in the expected range, which is [T‐1, T+11, where T is a new parameter smaller than R. Not expected intervals are handled differently to be corrected and in a few cases are eliminated. Results: Two types of tests were carried out. One uses real recordings containing artifacts. The other test was carried out by means of a computer program that generates corrupted data. R value was set in 0.2 and T in 0.05. Comparisons between our filter and other automatic algorithms were done. Conclusions: Our filter is fast and strong enough to perform an accurate automatic correction.