Paper 1: Cardiotocographic Diagnosis of Fetal Health based on Multiclass Morphologic Pattern Predictions using Deep Learning Classification
Abstract: Medical complications of pregnancy and pregnancy-related deaths continue to remain a major global challenge today. Internationally, about 830 maternal deaths occur every day due to pregnancy-related or childbirth-related complications. In fact, almost 99% of all maternal deaths occur in developing countries. In this research, an alternative and enhanced artificial intelligence approach is proposed for cardiotocographic diagnosis of fetal assessment based on multiclass morphologic pattern predictions, including 10 target classes with imbalanced samples, using deep learning classification models. The developed model is used to distinguish and classify the presence or absence of multiclass morphologic patterns for outcome predictions of complications during pregnancy. The testing results showed that the developed deep neural network model achieved an accuracy of 88.02%, a recall of 84.30%, a precision of 85.01%, and an F-score of 0.8508 in average. Thus, the developed model can provide highly accurate and consistent diagnoses for fetal assessment regarding complications during pregnancy, thereby preventing and/or reducing fetal mortality rate as well as maternal mortality rate during and following pregnancy and childbirth, especially in low-resource settings and developing countries.
Keywords: Activation function; deep learning; deep neural network; dropout; ensemble learning; multiclass; regularization; cardiotocography; complications during pregnancy; fetal heart rate