A new neural network approach for estimating fetal weight

We propose interactive information system based on experimental data description by means of a multivariate Gaussian probability model designed using artificial neural networks to improve accuracy of fetal weight estimation from routine ultrasound biometric parameters. The proposed system monitors and corrects on-line errors in gestational age estimation and biometrical measurements by ultrasounds. Well-controlled simulation experiments led to substantial improvement in the prediction of birthweight using ultrasonic examination of the fetus in the last week of pregnancy, thus suggesting a reliable use in clinical obstetrics management.