A New Model for Birth Weight Prediction Using 2- and 3-Dimensional Ultrasonography by Principal Component Analysis: A Chinese Population Study

Analysis of Publication Activity and Research Trends in the Field of AI Medical Applications: Network Approach

Artificial intelligence (AI) has revolutionized numerous industries, including medicine. In recent years, the integration of AI into medical practices has shown great promise in enhancing the accuracy and efficiency of diagnosing diseases, predicting patient outcomes, and personalizing treatment plans. This paper aims at the exploration of the AI-based medicine research using network approach and analysis of existing trends based on PubMed. Our findings are based on the results of PubMed search queries and analysis of the number of papers obtained by the different search queries. Our goal is to explore how are the AI-based methods used in healthcare research, which approaches and techniques are the most popular, and to discuss the potential reasoning behind the obtained results. Using analysis of the co-occurrence network constructed using VOSviewer software, we detected the main clusters of interest in AI-based healthcare research. Then, we proceeded with the thorough analysis of publication activity in various categories of medical AI research, including research on different AI-based methods applied to different types of medical data. We analyzed the results of query processing in the PubMed database over the past 5 years obtained via a specifically designed strategy for generating search queries based on the thorough selection of keywords from different categories of interest. We provide a comprehensive analysis of existing applications of AI-based methods to medical data of different modalities, including the context of various medical fields and specific diseases that carry the greatest danger to the human population.

Performance of whole-genome promoter nucleosome profiling of maternal plasma cell-free DNA for prenatal noninvasive prediction of fetal macrosomia: a retrospective nested case-control study in mainland China

Background

Fetal macrosomia is common occurrence in pregnancy, which is associated with several adverse prognosis both of maternal and neonatal. While, the accuracy of prediction of fetal macrosomia is poor. The aim of this study was to develop a reliable noninvasive prediction classifier of fetal macrosomia.

Methods

A total of 3600 samples of routine noninvasive prenatal testing (NIPT) data at 12^+ 0–27^+ 6 weeks of gestation, which were subjected to low-coverage whole-genome sequencing of maternal plasma cell-free DNA (cfDNA), were collected from three independent hospitals. We identified set of genes with significant differential coverages by comparing the promoter profiling between macrosomia cases and controls. We selected genes to develop classifier for noninvasive predicting, by using support vector machine (SVM) and logistic regression models, respectively. The performance of each classifier was evaluated by area under the curve (AUC) analysis.

Results

According to the available follow-up results, 162 fetal macrosomia pregnancies and 648 matched controls were included. A total of 1086 genes with significantly differential promoter profiling were found between pregnancies with macrosomia and controls (p < 0.05). With the AUC as a reference，the classifier based on SVM (C_MA-A2) had the best performance, with an AUC of 0.8256 (95% CI: 0.7927–0.8586).

Conclusions

Our study provides that assessing the risk of fetal macrosomia by whole-genome promoter nucleosome profiling of maternal plasma cfDNA based on low-coverage next-generation sequencing is feasible.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12884-022-05027-w.

Birth weight prediction models for the different gestational age stages in a Chinese population

The study aims to develop new birth weight prediction models for different gestational age stages using 2-dimensional (2D) ultrasound measurements in a Chinese population. 2D ultrasound was examined in pregnant women with normal singleton within 3 days prior to delivery (28–42 weeks’ gestation). A total of 19,310 fetuses were included in the study and randomly split into the training group and the validation group. Gestational age was divided into five stages: 28–30, 31–33, 34–36, 37–39 and 40–42 weeks. Multiple linear regression (MLR), fractional polynomial regression (FPR) and volume-based model (VM) were used to develop birth weight prediction model. New staged prediction models (VM for 28–36 weeks, MLR for 37–39 weeks, and FPR for 40–42 weeks) provided lower systematic errors and random errors than previously published models for each gestational age stage in the training group. The similar results were observed in the validation group. Compared to the previously published models, new staged models had the lowest aggregate systematic error (0.31%) and at least a 19.35% decrease; at least a 4.67% decrease for the root-mean-square error (RMSE). The prediction rates within 5% and 10% of birth weight for new staged models were higher than those for previously published models, which were 54.47% and 85.10%, respectively. New staged birth weight prediction models could improve the accuracy of birth weight estimation for different gestational age stages in a Chinese population.