Cardiotocography (CTG) is a primary tool for real-time monitoring fetal heart rate (FHR) and uterine contraction (UC) signals during labor. Nowadays CTG signal classification models based on deep learning have facilitated end-to-end CTG interpretation through deep neural networks for automatic feature extraction. However, the extreme rarity of abnormal cases in clinical settings has posed great challenges for CTG deep learning. Meanwhile, most Generative Adversarial Network (GAN)-based studies focused on FHR signal generation and ignored the importance of UC signals during labor. Hence, we introduced CTGGAN, a modified GAN specifically designed to simulate both FHR and UC intrapartum signals simultaneously. Our approach incorporated Wasserstein Distance to prevent gradient vanishing during training, alongside a set of latent code vectors that enhance the generator controllability. We also integrated an auxiliary classifier to impose categorical constraints, thereby enriching the diversity of the generated samples. The experimental results demonstrated that CTGGAN effectively generated CTG signals that closely resembled real CTG signals. In addition, the incorporation of CTGGAN significantly improved performance, with a 19.18% increase in sensitivity over the traditional classification without GAN. Furthermore, the comparison with several state-of-the-art CTG classification models underscored the discriminative performance of our CTGGAN. In conclusion, this study effectively enhanced the detection of abnormal cases for intrapartum fetal monitoring.

Fine spectrum refines conventional spectral sounding channels by increasing the spectral resolution to improve further the vertical resolution for accurate detection of atmospheric parameters. The 50- to 60-GHz absorption band is mainly used to sound the vertical distribution of atmospheric temperature in the troposphere and stratosphere, and 53 GHz is located in the weak oxygen absorption line. FengYun-3E (FY-3E) Microwave Temperature Sounder-III (MWTS-III) has two sounding channels added to this absorption band, and initially realized the refinement of the sampling channels. In this article, by establishing the assimilation operator for the radiance data of MWTS-III in China Meteorological Administration Global Assimilation Forecasting System (CMA-GFS) 4D-Var, and using the in-orbit observations from FY-3E MWTS-III, we investigate the impact of refining sounding channels at 53 GHz on numerical weather prediction (NWP). Results show that the temperature increment has the largest value and the widest influencing area near the peak weighting functions of channels, and the intensity and range of the temperature increment also change as the number of assimilated channels increases. An increment index can help quantify the impact of channel refinement on the forecast field. After single-cycle assimilation, the root mean square error (RMSE) of atmospheric parameters in the analysis field shows a decreasing trend. After continuous assimilation, the RMSE of geopotential height has a positive effect on short-term forecasts as the number of assimilated channels increases. It can be seen that channel refinement at 53 GHz positively impacts the performance of NWP.

During intrapartum fetal monitoring, it is significant for early detection and diagnosis of fetal distress. However, the traditional cardiotocography (CTG) interpretation methods heavily rely on physicians’ experience and lack consideration of the clinical features of pregnant women. To overcome these challenges, we propose a multimodal deep learning approach using Convolutional Neural Networks (CNN) and Vision Transformer (ViT) to end-to-end extract detailed and global deep features of CTG signal, respectively, and fuse the clinical features of pregnant women to predict fetal status. The experimental results demonstrate that the combination of CNN and ViT achieves outstanding performance with an average F1 value of 0.74 and an AUC value of 0.84. Furthermore, incorporating clinical features of pregnant women improves the model’s performance with an average F1 value of 0.78 and an area under the curve (AUC) value of 0.87. In summary, the proposed multimodal deep learning model shows the feasibility and effectiveness for intrapartum fetal monitoring.

Late fetal growth restriction (FGR) is a common complication of pregnancy characterized by chronic hypoxia. However, late FGR is in a dilemma of the high incidence but low detection rate. Depending on the non-invasiveness and convenient operation, the routine cardiotocography (CTG) allows continuous monitoring fetal heart rate (FHR) to assess fetal intrauterine stockpiling ability. In this paper, we aimed to explore the FHR pattern of late FGR in routine CTG. For analysis, the FHR features were acquired using routine CTG in a population of 160 healthy and 102 late FGR fetuses published in IEEE Dataport. First, we explored the relationships among FHR features and their importance on late FGR assessment by utilizing hypothesis testing, principal component analysis (PCA) and Spearman correlation analysis. Second, we presented a regression coefficient-based backward-stepwise-selection of association rules analysis (ARA) called backward-stepwise Max-R2Apriori ARA, to find the optimum itemset that helps diagnose late FGRs from healthy fetuses. The hypothesis testing, PCA and Spearman correlation analysis found eight FHR features were highly relevant to the late FGR. Moreover, the backward-stepwise Max-R2 Apriori ARA validated the correlation and interpretation about FHR features of late FGR. In conclusion, the analysis results are consistent with clinical knowledge on late FGR and help screen late FGR in antepartum fetal monitoring.

Cardiotocography (CTG) monitoring is a primary tool to assess the health of the fetus. It is widely used to identify the risk of fetal distress. With the outbreak of big data and artificial intelligence, the use of machine learning to assist obstetricians in CTG interpretation is important to improve diagnostic accuracy and save medical resources. However, imbalanced CTG data brings great challenges to machine learning in intelligent multi-classification. Therefore, we propose a stacked model based on Extreme-Gradient Boosting and Random Forest (XGB-RF) feature selection. Firstly, we use the XGB-RF feature selection method to extract the CTG features of significant influence. Then, we implement the fusion idea of stacking to construct the intelligent model that has a strong anti-interference ability under unbalanced CTG data with selected diverse classifiers. The experimental results showed that compared with existing CTG classification models in the public CTG dataset, the proposed model has further improved the performance and reduced the misjudgment between different classes. According to the results, the accuracy is 96.08%, the F1 score is 93.36%, and area under the ROC curve (AUC) is 0.9883. This indicates that the proposed techniques have a great clinical significance in fetal health monitoring.

The realization of intelligent fetal monitoring is helpful to timely detect fetal abnormality and save medical costs. However, the probability of misjudgment tends to be high when modeling the imbalanced clinical data directly. Moreover, reliable interpretation of cardiotocography (CTG) requires multiple obstetricians to annotate, which consumes excessive time and labor cost. In this paper, a K-means method based on adaptive Genetic Algorithm-Fl balanced diversity Weighted Query-by-Committee (QBC) algorithm (KGA-WQBC) was proposed to solve the above problems. The proposed active learning algorithm was improved from the following two aspects. Firstly, an adaptive KGA operator was designed to conduct a preliminary selection of a large number of unlabeled samples, which increased the accuracy of the committee. Then, the WQBC operator was designed to improve the measurement of the committee's disagreement degree after data exploration. Compared with other initial sample selection strategies and disagreement degree weighting means, the results showed that only 41% of the labeled instances were used to make the evaluation index as high as 98.02%, which had the best overall performance. In conclusion, the proposed KGA-WQBC algorithm is reasonable and feasible, and effectively solves the problem of imbalanced data and reduces the cost of medical personnel labeling for intelligent fetal monitoring.