Wavelets and the Unborn Child.

During labour, the attending medical staff use fetal heart rate recordings for evaluation of fetal well being and may base immediate intervention, such as a Caesarean section or taking a fetal scalp blood sample, on this. Using characteristics derived in real-time from the heart rate, obstetricians can predict a good outcome very well. However, in cases of fetal heart rate patterns considered `bad' by the obstetrician, at least half of these turn out to have been false alarms and the (operative) intervention unnecessary. Decision making can be improved by providing relevant information contained in the heart rate on a more solid, objective basis, making it independent of the personal experience of the specialist. This is enabled by recent progress in the modelling and analysis of heartbeat inter-beat dynamics, using the most advanced methods of signal processing (wavelet transform). CWI is tackling the mathematical side of this problem in cooperation with the Academic Medical Centre in Amsterdam (W.J. van Wijngaarden) and the Institute of Information and Computing Sciences of Utrecht University (R. Castelo). After mimicing the obstetrician's expert knowledge, the ultimate goal is to provide better than human performance by automated learning of predictive models.