Detection and interpolation of outliers in biosignals.

Detection of outliers in recorded biosignals and their interpolation is described. The detection of outliers is based on the simple rule of 3 sigma. The subsequent interpolation is performed according to the method proposed by Brubacher and Wilson for autoregressive processes. The output of the suggested procedure can be treated by standard programs of statistical analysis as long as the influence of outliers is eliminated. The program in Fortran IV developed in the Institute of Hygiene and Epidemiology, Prague, performs automatically the detection of outliers in an input signal and their replacement by means of the proposed interpolation method.