Fractal characteristics of breath to breath timing in sleeping infants.

We examined interbreath interval (IBI) time series of 19 term infants during active and quiet sleep for fractal properties using Fano factor analysis. For each time series we calculated the fractal exponent (alpha), comparing alpha for the original time series with two forms of surrogate data, a temporally independent surrogate set and an autoregressive surrogate set. alpha values were normally distributed between 0.79 and -0.22, and did not differ with sleep state. The fractal characteristics of the original time series were not retained in the temporally independent surrogate time series indicating that the distribution of intervals alone was not fractal, but were retained using autoregressive surrogates with an order of 10, suggesting that the fractal properties of the IBI time series were related to correlations between successive breaths. These observations suggest that some of the respiratory variability that occurs during sleep in infants, which in the past has been regarded as stochastic noise, may be the product of deterministic processes.