One of the advantages of signal analysis is that it can be applied to an enormous range of problems. One such problem is the question of when clinicians at a child-birth should intervene to prevent damage to new-born infants. This work, done in collaboration with the Nuffield Department of Obstetrics and Gynaecology, is aimed at developing an automated system to detect abnormalities in foetal recordings, routinely recorded during labour. This would assist the clinician in their decision making of whether or not to intervene. Dealing with extremely noisy signals with lots of artefacts is a hugely difficult problem, so lots of recordings are needed as well as some very clever analysis. Using the world's largest database, we are working on how to relate new features extracted from signals to clinical outcome as robustly as possible to allow clinicians to decide to intervene or not in individual births.