Research Outputs
Simple and fast prediction of gestational diabetes mellitus based on machine learning and near-infrared spectra of serum: A proof of concept study at different stages of pregnancy
Gestational diabetes mellitus (GDM) is a hyperglycemic state that is typically diagnosed by an oral glucose tolerance test (OGTT), which is unpleasant, time-consuming, has low reproducibility, and results are tardy. The machine learning (ML) predictive models that have been proposed to improve GDM diagnosis are usually based on instrumental methods that take hours to produce a result. Near-infrared (NIR) spectroscopy is a simple, fast, and low-cost analytical technique that has never been assessed for the prediction of GDM. This study aims to develop ML predictive models for GDM based on NIR spectroscopy, and to evaluate their potential as early detection or alternative screening tools according to their predictive power and duration of analysis. Serum samples from the first trimester (before GDM diagnosis) and the second trimester (at the time of GDM diagnosis) of pregnancy were analyzed by NIR spectroscopy. Four spectral ranges were considered, and 80 mathematical pretreatments were tested for each. NIR data-based models were built with single- and multi-block ML techniques. Every model was subjected to double cross-validation. The best models for first and second trimester achieved areas under the receiver operating characteristic curve of 0.5768 ± 0.0635 and 0.8836 ± 0.0259, respectively. This is the first study reporting NIR-spectroscopy-based methods for the prediction of GDM. The developed methods allow for prediction of GDM from 10 µL of serum in only 32 min. They are simple, fast, and have a great potential for application in clinical practice, especially as alternative screening tools to the OGTT for GDM diagnosis.
Pathophysiological Role of Genetic Factors Associated With Gestational Diabetes Mellitus
Gestational Diabetes Mellitus (GDM) is a highly prevalent maternal pathology characterized by maternal glucose intolerance during pregnancy that is, associated with severe complications for both mother and offspring. Several risk factors have been related to GDM; one of the most important among them is genetic predisposition. Numerous single nucleotide polymorphisms (SNPs) in genes that act at different levels on various tissues, could cause changes in the expression levels and activity of proteins, which result in glucose and insulin metabolism dysfunction. In this review, we describe various SNPs; which according to literature, increase the risk of developing GDM. These SNPs include: (1) those associated with transcription factors that regulate insulin production and excretion, such as rs7903146 (TCF7L2) and rs5015480 (HHEX); (2) others that cause a decrease in protective hormones against insulin resistance such as rs2241766 (ADIPOQ) and rs6257 (SHBG); (3) SNPs that cause modifications in membrane proteins, generating dysfunction in insulin signaling or cell transport in the case of rs5443 (GNB3) and rs2237892 (KCNQ1); (4) those associated with enzymes such as rs225014 (DIO2) and rs9939609 (FTO) which cause an impaired metabolism, resulting in an insulin resistance state; and (5) other polymorphisms, those are associated with growth factors such as rs2146323 (VEGFA) and rs755622 (MIF) which could cause changes in the expression levels of these proteins, producing endothelial dysfunction and an increase of pro-inflammatory cytokines, characteristic on GDM. While the pathophysiological mechanism is unclear, this review describes various potential effects of these polymorphisms on the predisposition to develop GDM.
Machine learning-based models for gestational diabetes mellitus prediction before 24–28 weeks of pregnancy: A review
Gestational Diabetes Mellitus (GDM) is a hyperglycemia state that impairs maternal and offspring health, short and long-term. It is usually diagnosed at 24–28 weeks of pregnancy (WP), but at that time the fetal phenotype is already altered. Machine learning (ML)-based models have emerged as an auspicious alternative to predict this pathology earlier, however, they must be validated in different populations before their implementation in routine clinical practice. This review aims to give an overview of the ML-based models that have been proposed to predict GDM before 24–28 WP, with special emphasis on their current validation state and predictive performance. Articles were searched in PubMed. Manuscripts written in English and published before January 1, 2022, were considered. 109 original research studies were selected, and categorized according to the type of variables that their models involved: medical, i.e. clinical and/or biochemical parameters; alternative, i.e. metabolites, peptides or proteins, micro-ribonucleic acid molecules, microbiota genera, or other variables that did not fit into the first category; or mixed, i.e. both medical and alternative data. Only 8.3 % of the reviewed models have had validation in independent studies, with low or moderate performance for GDM prediction. In contrast, several models that lack of independent validation have shown a very high predictive power. The evaluation of these promising models in future independent validation studies would allow to assess their performance on different populations, and continue their way towards clinical implementation. Once settled, ML-based models would help to predict GDM earlier, initiate its treatment timely and prevent its negative consequences on maternal and offspring health.
Maternal thyroid profile in first and second trimester of pregnancy is correlated with gestational diabetes mellitus through machine learning
There is evidence about a possible relationship between thyroid abnormalities and gestational diabetes mellitus (GDM). However, there is still no conclusive data on this dependence, since no strong correlation has been proved. In this work, we used machine learning to determine whether there is a correlation between maternal thyroid profile in first and second trimester of pregnancy and GDM. Using principal component analysis, it was possible to find an evident correlation between both, which could be used as a complement for a more sensitive GDM diagnosis.