A comparative study of usefulness of quantitative parameters in the description of cardiotocogram records

This study was an attempt to choose the optimized set of the features obtained from cardiotocogram records, which could explain in the best way the variability of foetal acid-base status parameters at labour. We prepared a package of the procedures in FORTRAN language which provides the possibilities for automatic detection and recognition of the long-term variability such as accelerations and decelerations and assessing some parameters connected with short-term variability (among other things oscillation bandwidth and number of baseline crossings). The acquired variables were subjected to multivariate statistical analysis to estimate the predictive power of a particular parameter. We used the described method to predict the variation of the foetal scalp blood pH and base deficit. This let us construct a set of variables which explains the maximal part of the variability of gasometric parameters. It affords possibilities for the estimation of the relative power of a particular component calculated from the cardiotocogram pattern. We obtain 41.7% of the explained variance for pH and 38.3% for base deficit.