Dimension estimates and physiological data

Dimension estimates for data from physiological systems are notoriously difficult since the data are far from ideal in the sense of deterministic dynamical systems. Possible pitfalls and necessary precautions are pointed out and a recipe is given which is viable for those researchers who want to use the Grassberger-Procaccia algorithm but who are not familiar with the vast existing literature on dimension estimates. The relevance of dimension estimates for the characterization of physiological data is discussed, where both the cases of finding and not finding a low dimension are considered. (c) 1995 American Institute of Physics.