Predicting preterm birth using explainable machine learning in a prospective cohort of nulliparous and multiparous pregnant women.

Preterm birth (PTB) presents a complex challenge in pregnancy, often leading to significant perinatal and long-term morbidities. "While machine learning (ML) algorithms have shown promise in PTB prediction, the lack of interpretability in existing models hinders their clinical utility. This study aimed to predict PTB in a pregnant population using ML models, identify the key risk factors associated with PTB through the SHapley Additive exPlanations (SHAP) algorithm, and provide comprehensive explanations for these predictions to assist clinicians in providing appropriate care. This study analyzed a dataset of 3509 pregnant women in the United Arab Emirates and selected 35 risk factors associated with PTB based on the existing medical and artificial intelligence literature. Six ML algorithms were tested, wherein the XGBoost model exhibited the best performance, with an area under the operator receiving curves of 0.735 and 0.723 for parous and nulliparous women, respectively. The SHAP feature attribution framework was employed to identify the most significant risk factors linked to PTB. Additionally, individual patient analysis was performed using the SHAP and the local interpretable model-agnostic explanation algorithms (LIME). The overall incidence of PTB was 11.23% (11 and 12.1% in parous and nulliparous women, respectively). The main risk factors associated with PTB in parous women are previous PTB, previous cesarean section, preeclampsia during pregnancy, and maternal age. In nulliparous women, body mass index at delivery, maternal age, and the presence of amniotic infection were the most relevant risk factors. The trained ML prediction model developed in this study holds promise as a valuable screening tool for predicting PTB within this specific population. Furthermore, SHAP and LIME analyses can assist clinicians in understanding the individualized impact of each risk factor on their patients and provide appropriate care to reduce morbidity and mortality related to PTB.

Node embedding-based graph autoencoder outlier detection for adverse pregnancy outcomes.

Adverse pregnancy outcomes, such as low birth weight (LBW) and preterm birth (PTB), can have serious consequences for both the mother and infant. Early prediction of such outcomes is important for their prevention. Previous studies using traditional machine learning (ML) models for predicting PTB and LBW have encountered two important limitations: extreme class imbalance in medical datasets and the inability to account for complex relational structures between entities. To address these limitations, we propose a node embedding-based graph outlier detection algorithm to predict adverse pregnancy outcomes. We developed a knowledge graph using a well-curated representative dataset of the Emirati population and two node embedding algorithms. The graph autoencoder (GAE) was trained by applying a combination of original risk factors and node embedding features. Samples that were difficult to reconstruct at the output of GAE were identified as outliers considered representing PTB and LBW samples. Our experiments using LBW, PTB, and very PTB datasets demonstrated that incorporating node embedding considerably improved performance, achieving a 12% higher AUC-ROC compared to traditional GAE. Our study demonstrates the effectiveness of node embedding and graph outlier detection in improving the prediction performance of adverse pregnancy outcomes in well-curated population datasets.

LecturePlus: a learner-centered teaching method to promote deep learning.

A new teaching format, the LecturePlus, was formulated as a lecture followed by small-group learning activities. This study assessed the effectiveness of LecturePlus in medical education. An interventional study was conducted among final-year medical students, with 74 students in the experimental group and 96 students in the control group. Each LecturePlus lasted ∼1 h and was conducted with 12-18 students. A LecturePlus session comprised of a lecture followed by small-group problem-solving exercises. The exercises were clinical cases with open-ended short-answer questions. Students were divided into groups of three (triads) for these exercises. A faculty tutor assisted the small groups as needed. Closure was achieved through a discussion moderated by the tutor. Learning outcomes were assessed via the National Board of Medical Examiners (NBME) subject scores and compared with those of the preceding academic year. An additional multiple-choice question (MCQ) test was administered before and after the clerkship. The MCQ test showed improvement in knowledge application (P < 0.001, partial eta squared = 0.42). There was a statistically significant improvement in adjusted NBME scores among female students (74.8 vs. 71.8; P = 0.017) but not among male students. An anonymous written questionnaire survey showed high ratings for LecturePlus (95% selecting yes or partly yes to overall satisfaction). LecturePlus is an instructional strategy that integrates a lecture with learning activities. It can be scaled to large class sizes facilitated by one teacher. By combining didactic teaching with problem-solving, this new instructional strategy may foster application of knowledge.NEW & NOTEWORTHY We developed a new structured teaching format, the LecturePlus, to promote deep learning. A LecturePlus session consists of a brief lecture, followed by small-group problem-solving exercises, ending with a closing discussion moderated by the teacher. During the small-group exercises, students were divided into groups of three (triads) and given case-based problems. One faculty tutor supervised the entire session. LecturePlus resulted in improved learning outcomes and was rated highly by medical students.

Infant Low Birth Weight Prediction Using Graph Embedding Features.

Low Birth weight (LBW) infants pose a serious public health concern worldwide in both the short and long term for infants and their mothers. Infant weight prediction prior to birth can help to identify risk factors and reduce the risk of infant morbidity and mortality. Although many Machine Learning (ML) algorithms have been proposed for LBW prediction using maternal features and produced considerable model performance, their performance needs to be improved so that they can be adapted in real-world clinical settings. Existing algorithms used for LBW classification often fail to capture structural information from the tabular dataset of patients with different complications. Therefore, to improve the LBW classification performance, we propose a solution by transforming the tabular data into a knowledge graph with the aim that patients from the same class (normal or LBW) exhibit similar patterns in the graphs. To achieve this, several features related to each node are extracted such as node embedding using node2vec algorithm, node degree, node similarity, nearest neighbors, etc. Our method is evaluated on a real-life dataset obtained from a large cohort study in the United Arab Emirates which contains data from 3453 patients. Multiple experiments were performed using the seven most commonly used ML models on the original dataset, graph features, and a combination of features, respectively. Experimental results show that our proposed method achieved the best performance with an area under the curve of 0.834 which is over 6% improvement compared to using the original risk factors without transforming them into knowledge graphs. Furthermore, we provide the clinical relevance of the proposed model that are important for the model to be adapted in clinical settings.

Vitamin A deficiency in rats induces anatomic and metabolic changes comparable with those of neurodegenerative disorders.

Anatomic and metabolic changes in central nervous system induced by 14 wk of vitamin A deprivation (VAD) were monitored and quantified in rats. In vivo brain magnetic resonance imaging (4.7T) was performed at 5, 7, 9, 11, and 14 wk of each diet after weaning in the following: 1) VAD group; 2) control pair-fed group; and 3) control group that consumed the diet ad libitum (1.15 microg retinol/g diet). After 14 wk, high-resolution magic angle spinning proton NMR spectroscopy (11.7T) was performed on small samples of cortex, hippocampus, and striatum. Serum retinol concentrations remained stable and cerebral volume (CV) increased as a linear function of body weight in the ad libitum group (R(2) = 0.78; P = 0.047) and pair-fed controls (R(2) = 0.78; P = 0.046). In VAD rats, retinol decreased from the onset of deprivation (2.2 +/- 0.14 micromol/L) to reach 0.3 +/- 0.13 micromol/L at wk 5, followed by a stopping of body weight gain from wk 7. In VAD rats, the CV decreased from wk 5 and reached a value 11% lower than that of the control group (P < 0.001) at wk 14 and was correlated with retinol status (R(2) = 0.99; P = 0.002). The VAD hippocampal volume decreased beginning at wk 9 and was 22% lower than that of the control group at wk 14 (P < 0.001). Compared with the control, VAD led to lower N acetyl aspartate:creatine+phosphocreatine (Cr) in cortex (-36%), striatum (-22%), and hippocampus (-19%) and higher myoinositol:Cr in cortex (+127%) and striatum (+150%). VAD induced anatomic and metabolic changes comparable to those associated with neurodegenerative disorders. By wk 7 of deprivation, the slowing in cerebral growth that correlated with the retinol level could be considered as a predictive marker of brain disorders, confirmed by metabolic data from VAD rats after 14 wk.

Retinoic acid normalizes nuclear receptor mediated hypo-expression of proteins involved in beta-amyloid deposits in the cerebral cortex of vitamin A deprived rats.

Recent data have revealed that disruption of vitamin A signaling observed in Alzheimer's disease (AD) leads to a deposition of beta-amyloid (Abeta). The aim of this study was to precise the role of vitamin A and its nuclear receptors (RAR) in the processes leading to the Abeta deposits. Thus, the effect of vitamin A depletion and subsequent administration of retinoic acid (RA, the active metabolite of vitamin A) on the expression of RARbeta, and of proteins involved in amyloidogenic pathway, e.g., amyloid precursor protein (APP), beta-secretase enzyme (BACE), and APP carboxy-terminal fragment (APP-CTF) was examined in the whole brain, hippocampus, striatum, and cerebral cortex of rats. Rats fed a vitamin A-deprived diet for 13 weeks exhibited decreased amount of RARbeta, APP695, BACE, and of APP-CTF in the whole brain and in the cerebral cortex. Administration of RA is able to restore all expression. The results suggest that fine regulation of vitamin A mediated gene expression seems fundamental for the regulation of APP processing.