CDT Postgraduate Studentship in Healthcare Innovation (RCUK Digital Economy Programme grant number EP/G036861/1)
Time series analysis of the fetal heart rate
The inherent difficulty of access to fetus limits the clinicians’ capability to assess the fetal condition during labour. The most common measurements taken are the fetal heart rate (FHR) and maternal uterine activity (UA), which have been used successfully to identify abnormalities in the early stages of labour. Now the Oxford Centre for Fetal Monitoring Technologies (OCFMT) is developing a system to detect abnormalities in the final stages of labour, a task made more difficult by the stresses induced as a normal part of birth. The Oxford System (OxSys) extracts useful information from the FHR and UA signals, which is used to classify traces.
I have focused on determining the importance of the evolution of these features over time, using Hidden Markov Models to construct an optimal ‘gray-box’ classifier, from which information about the normal and abnormal progression of labour can be discovered. The model has demonstrated the improved performance of classification using time series of features, and hinted at the increased separability of normal and abnormal cases in the early portion of second stage labour.