Assessment of Fetal Wellbeing via a Novel Neural Network

D.P. Scarpazza (Italy), M.H. Graupe, D. Graupe, and C.J. Hobel (USA)


fetal distress, neural network, signalprocessing, diagnosis.


Electronic fetal heart rate monitoring is now a commonly used method to aid clinicians in their surveillance of fetal health. Nevertheless visual interpretation of data is still subject to controversies, lack of commonly-accepted definitions and automated algorithms. We propose a set of formalizations and algorithms, based upon digital signal processing and on a novel neural network which should provide a flexible tool, capable of acquiring patients’ data and of formulating a reliable assessment on the fetal health. Furthermore, we discuss a neural network-based diagnosis system and a pre-processing subsystem at the neural network’s input, for data reduction. The neural network employed is specifically suited for rapid analysis of huge data sets and categories as required for the present problem. The present study is preliminary and is based on only a limited set of training and validation data sets (patterns). Still, the results shown by the neural network appears rather encouraging, especially in terms of sensitivity to positive case and a study on a much larger data base is under way.

Important Links:

Go Back