Fractal structure underlying patterns of premature heart beats

Distinctive patterns of premature heart beats are frequently observed in clinical data, yet transitions from one pattern to another often appear to occur unpredictably. In this study we investigate the origins of this apparent lack of long-term order by processing the time intervals between successive R wave features of the electrocardiogram to obtain a symbolic description of all characteristic patterns of premature beats observed in a clinical data set. These characteristic patterns are found to be 1) associated with fixed ranges of heart rate over periods of many hours and 2) organized into an interwoven fractal structure that corresponds to the sequence of mode-locked states predicted by a nonlinear mathematical model. The apparent randomness of transitions from one pattern to another results from the normal fluctuations in heart rate which can sample many such patterns on a minute-by-minute basis.