Real-Time Numerical Filtering of Physiological Signals

The minimization of the sampling rate necessary to store a signal frequently requires filtering of unwanted higher frequency components in order to avoid aliasing errors. In many physiological situations the desired cutoff frequency is too low to be realized by a practical passive analog filter. In such a case either an active analog filter may be used, or the signal may be initially sampled at a higher rate to avoid aliasing, then numerically filtered in real-time, and the resulting smoothed signal sampled at a lower rate (the effective sampling frequency). The purpose of this communication is to describe the use of a class of low-pass digital filters which we have found useful for this purpose.