Artificial intelligence and the analysis of multi-platform metabolomics data for the detection of intrauterine growth restriction

Prediction of adverse neonatal adaptation in fetuses with severe fetal growth restriction after 34 weeks of gestation

OBJECTIVE

To establish a predictive model for adverse immediate neonatal adaptation (INA) in fetuses with suspected severe fetal growth restriction (FGR) after 34 gestational weeks (GW).

METHODS

We conducted a retrospective observational study at the University Hospitals of Strasbourg between 2000 and 2020, including 1,220 women with a singleton pregnancy and suspicion of severe FGR who delivered from 34 GW. The primary outcome (composite) was INA defined as Apgar 5-minute score <7, arterial pH <7.10, immediate transfer to pediatrics, or the need for resuscitation at birth. We developed and tested a logistic regression predictive model.

RESULTS

Adverse INA occurred in 316 deliveries. The model included six features available before labor: parity, gestational age, diabetes, middle cerebral artery Doppler, cerebral-placental inversion, onset of labor. The model could predict individual risk of adverse INA with confidence interval at 95 %. Taking an optimal cutoff threshold of 32 %, performances were: sensitivity 66 %; specificity 83 %; positive and negative predictive values 60 % and 87 % respectively, and area under the curve 78 %.

DISCUSSION

The predictive model showed good performances and a proof of concept that INA could be predicted with pre-labor characteristics, and needs to be investigated further.

Lipid profile of circulating placental extracellular vesicles during pregnancy identifies foetal growth restriction risk

Small-for-gestational age (SGA) neonates exhibit increased perinatal morbidity and mortality, and a greater risk of developing chronic diseases in adulthood. Currently, no effective maternal blood-based screening methods for determining SGA risk are available. We used a high-resolution MS/MSALL shotgun lipidomic approach to explore the lipid profiles of small extracellular vesicles (sEV) released from the placenta into the circulation of pregnant individuals. Samples were acquired from 195 normal and 41 SGA pregnancies. Lipid profiles were determined serially across pregnancy. We identified specific lipid signatures of placental sEVs that define the trajectory of a normal pregnancy and their changes occurring in relation to maternal characteristics (parity and ethnicity) and birthweight centile. We constructed a multivariate model demonstrating that specific lipid features of circulating placental sEVs, particularly during early gestation, are highly predictive of SGA infants. Lipidomic-based biomarker development promises to improve the early detection of pregnancies at risk of developing SGA, an unmet clinical need in obstetrics.

Similarity network fusion to identify phenotypes of small-for-gestational-age fetuses

Fetal growth restriction (FGR) affects 5-10% of pregnancies, is the largest contributor to fetal death, and can have long-term consequences for the child. Implementation of a standard clinical classification system is hampered by the multiphenotypic spectrum of small fetuses with substantial differences in perinatal risks. Machine learning and multiomics data can potentially revolutionize clinical decision-making in FGR by identifying new phenotypes. Herein, we describe a cluster analysis of FGR based on an unbiased machine-learning method. Our results confirm the existence of two subtypes of human FGR with distinct molecular and clinical features based on multiomic analysis. In addition, we demonstrated that clusters generated by machine learning significantly outperform single data subtype analysis and biologically support the current clinical classification in predicting adverse maternal and neonatal outcomes. Our approach can aid in the refinement of clinical classification systems for FGR supported by molecular and clinical signatures.

Prediction Models for Intrauterine Growth Restriction Using Artificial Intelligence and Machine Learning: A Systematic Review and Meta-Analysis

BACKGROUND

IntraUterine Growth Restriction (IUGR) is a global public health concern and has major implications for neonatal health. The early diagnosis of this condition is crucial for obtaining positive outcomes for the newborn. In recent years Artificial intelligence (AI) and machine learning (ML) techniques are being used to identify risk factors and provide early prediction of IUGR. We performed a systematic review (SR) and meta-analysis (MA) aimed to evaluate the use and performance of AI/ML models in detecting fetuses at risk of IUGR.

METHODS

We conducted a systematic review according to the PRISMA checklist. We searched for studies in all the principal medical databases (MEDLINE, EMBASE, CINAHL, Scopus, Web of Science, and Cochrane). To assess the quality of the studies we used the JBI and CASP tools. We performed a meta-analysis of the diagnostic test accuracy, along with the calculation of the pooled principal measures.

RESULTS

We included 20 studies reporting the use of AI/ML models for the prediction of IUGR. Out of these, 10 studies were used for the quantitative meta-analysis. The most common input variable to predict IUGR was the fetal heart rate variability (n = 8, 40%), followed by the biochemical or biological markers (n = 5, 25%), DNA profiling data (n = 2, 10%), Doppler indices (n = 3, 15%), MRI data (n = 1, 5%), and physiological, clinical, or socioeconomic data (n = 1, 5%). Overall, we found that AI/ML techniques could be effective in predicting and identifying fetuses at risk for IUGR during pregnancy with the following pooled overall diagnostic performance: sensitivity = 0.84 (95% CI 0.80-0.88), specificity = 0.87 (95% CI 0.83-0.90), positive predictive value = 0.78 (95% CI 0.68-0.86), negative predictive value = 0.91 (95% CI 0.86-0.94) and diagnostic odds ratio = 30.97 (95% CI 19.34-49.59). In detail, the RF-SVM (Random Forest-Support Vector Machine) model (with 97% accuracy) showed the best results in predicting IUGR from FHR parameters derived from CTG.

CONCLUSIONS

our findings showed that AI/ML could be part of a more accurate and cost-effective screening method for IUGR and be of help in optimizing pregnancy outcomes. However, before the introduction into clinical daily practice, an appropriate algorithmic improvement and refinement is needed, and the importance of quality assessment and uniform diagnostic criteria should be further emphasized.

Precision Medicine Approaches with Metabolomics and Artificial Intelligence

Recent technological innovations in the field of mass spectrometry have supported the use of metabolomics analysis for precision medicine. This growth has been allowed also by the application of algorithms to data analysis, including multivariate and machine learning methods, which are fundamental to managing large number of variables and samples. In the present review, we reported and discussed the application of artificial intelligence (AI) strategies for metabolomics data analysis. Particularly, we focused on widely used non-linear machine learning classifiers, such as ANN, random forest, and support vector machine (SVM) algorithms. A discussion of recent studies and research focused on disease classification, biomarker identification and early diagnosis is presented. Challenges in the implementation of metabolomics–AI systems, limitations thereof and recent tools were also discussed.

The Exploration of Fetal Growth Restriction Based on Metabolomics: A Systematic Review

Fetal growth restriction (FGR) is a common complication of pregnancy and a significant cause of neonatal morbidity and mortality. The adverse effects of FGR can last throughout the entire lifespan and increase the risks of various diseases in adulthood. However, the etiology and pathogenesis of FGR remain unclear. This study comprehensively reviewed metabolomics studies related with FGR in pregnancy to identify potential metabolic biomarkers and pathways. Relevant articles were searched through two online databases (PubMed and Web of Science) from January 2000 to July 2022. The reported metabolites were systematically compared. Pathway analysis was conducted through the online MetaboAnalyst 5.0 software. For humans, a total of 10 neonatal and 14 maternal studies were included in this review. Several amino acids, such as alanine, valine, and isoleucine, were high frequency metabolites in both neonatal and maternal studies. Meanwhile, several pathways were suggested to be involved in the development of FGR, such as arginine biosynthesis, arginine, and proline metabolism, glyoxylate and dicarboxylate metabolism, and alanine, aspartate, and glutamate metabolism. In addition, we also included 8 animal model studies, in which three frequently reported metabolites (glutamine, phenylalanine, and proline) were also present in human studies. In general, this study summarized several metabolites and metabolic pathways which may help us to better understand the underlying metabolic mechanisms of FGR.

Metabolomic profiling of intrauterine growth-restricted preterm infants: a matched case-control study

BACKGROUND

The biochemical variations occurring in intrauterine growth restriction (IUGR), when a fetus is unable to achieve its genetically determined potential, are not fully understood. The aim of this study is to compare the urinary metabolomic profile between IUGR and non-IUGR very preterm infants to investigate the biochemical adaptations of neonates affected by early-onset-restricted intrauterine growth.

METHODS

Neonates born <32 weeks of gestation admitted to neonatal intensive care unit (NICU) were enrolled in this prospective matched case-control study. IUGR was diagnosed by an obstetric ultra-sonographer and all relevant clinical data during NICU stay were captured. For each subject, a urine sample was collected within 48 h of life and underwent untargeted metabolomic analysis using mass spectrometry ultra-performance liquid chromatography. Data were analyzed using multivariate and univariate statistical analyses.

RESULTS

Among 83 enrolled infants, 15 IUGR neonates were matched with 19 non-IUGR controls. Untargeted metabolomic revealed evident clustering of IUGR neonates versus controls showing derangements of pathways related to tryptophan and histidine metabolism and aminoacyl-tRNA and steroid hormones biosynthesis.

CONCLUSIONS

Neonates with IUGR showed a distinctive urinary metabolic profile at birth. Although results are preliminary, metabolomics is proving to be a promising tool to explore biochemical pathways involved in this disease.

IMPACT

Very preterm infants with intrauterine growth restriction (IUGR) have a distinctive urinary metabolic profile at birth. Metabolism of glucocorticoids, sexual hormones biosynthesis, tryptophan-kynurenine, and methionine-cysteine pathways seem to operate differently in this sub-group of neonates. This is the first metabolomic study investigating adaptations exclusively in extremely and very preterm infants affected by early-onset IUGR. New knowledge on metabolic derangements in IUGR may pave the ways to further, more tailored research from a perspective of personalized medicine.

Predicting Hypertension Subtypes with Machine Learning Using Targeted Metabolites and Their Ratios

Hypertension is a major global health problem with high prevalence and complex associated health risks. Primary hypertension (PHT) is most common and the reasons behind primary hypertension are largely unknown. Endocrine hypertension (EHT) is another complex form of hypertension with an estimated prevalence varying from 3 to 20% depending on the population studied. It occurs due to underlying conditions associated with hormonal excess mainly related to adrenal tumours and sub-categorised: primary aldosteronism (PA), Cushing’s syndrome (CS), pheochromocytoma or functional paraganglioma (PPGL). Endocrine hypertension is often misdiagnosed as primary hypertension, causing delays in treatment for the underlying condition, reduced quality of life, and costly antihypertensive treatment that is often ineffective. This study systematically used targeted metabolomics and high-throughput machine learning methods to predict the key biomarkers in classifying and distinguishing the various subtypes of endocrine and primary hypertension. The trained models successfully classified CS from PHT and EHT from PHT with 92% specificity on the test set. The most prominent targeted metabolites and metabolite ratios for hypertension identification for different disease comparisons were C18:1, C18:2, and Orn/Arg. Sex was identified as an important feature in CS vs. PHT classification.

A Metabolomic Profiling of Intra-Uterine Growth Restriction in Placenta and Cord Blood Points to an Impairment of Lipid and Energetic Metabolism

(1) Background: Intrauterine growth restriction (IUGR) involves metabolic changes that may be responsible for an increased risk of metabolic and cardiovascular diseases in adulthood. Several metabolomic profiles have been reported in maternal blood and urine, amniotic fluid, cord blood and newborn urine, but the placenta has been poorly studied so far. (2) Methods: To decipher the origin of this metabolic reprogramming, we conducted a targeted metabolomics study replicated in two cohorts of placenta and one cohort of cord blood by measuring 188 metabolites by mass spectrometry. (3) Results: OPLS-DA multivariate analyses enabled clear discriminations between IUGR and controls, with good predictive capabilities and low overfitting in the two placental cohorts and in cord blood. A signature of 25 discriminating metabolites shared by both placental cohorts was identified. This signature points to sharp impairment of lipid and mitochondrial metabolism with an increased reliance on the creatine-phosphocreatine system by IUGR placentas. Increased placental insulin resistance and significant alteration of fatty acids oxidation, together with relatively higher phospholipase activity in IUGR placentas, were also highlighted. (4) Conclusions: Our results show a deep lipid and energetic remodeling in IUGR placentas that may have a lasting effect on the fetal metabolism.

Artificial Intelligence and Circulating Cell-Free DNA Methylation Profiling: Mechanism and Detection of Alzheimer's Disease

Background: Despite extensive efforts, significant gaps remain in our understanding of Alzheimer’s disease (AD) pathophysiology. Novel approaches using circulating cell-free DNA (cfDNA) have the potential to revolutionize our understanding of neurodegenerative disorders. Methods: We performed DNA methylation profiling of cfDNA from AD patients and compared them to cognitively normal controls. Six Artificial Intelligence (AI) platforms were utilized for the diagnosis of AD while enrichment analysis was used to elucidate the pathogenesis of AD. Results: A total of 3684 CpGs were significantly (adj. p-value < 0.05) differentially methylated in AD versus controls. All six AI algorithms achieved high predictive accuracy (AUC = 0.949−0.998) in an independent test group. As an example, Deep Learning (DL) achieved an AUC (95% CI) = 0.99 (0.95−1.0), with 94.5% sensitivity and specificity. Conclusion: We describe numerous epigenetically altered genes which were previously reported to be differentially expressed in the brain of AD sufferers. Genes identified by AI to be the best predictors of AD were either known to be expressed in the brain or have been previously linked to AD. We highlight enrichment in the Calcium signaling pathway, Glutamatergic synapse, Hedgehog signaling pathway, Axon guidance and Olfactory transduction in AD sufferers. To the best of our knowledge, this is the first reported genome-wide DNA methylation study using cfDNA to detect AD.

Interpreting the role of nuchal fold for fetal growth restriction prediction using machine learning

The objective of the study is to investigate the effect of Nuchal Fold (NF) in predicting Fetal Growth Restriction (FGR) using machine learning (ML), to explain the model's results using model-agnostic interpretable techniques, and to compare the results with clinical guidelines. This study used second-trimester ultrasound biometry and Doppler velocimetry were used to construct six FGR (birthweight < 3rd centile) ML models. Interpretability analysis was conducted using Accumulated Local Effects (ALE) and Shapley Additive Explanations (SHAP). The results were compared with clinical guidelines based on the most optimal model. Support Vector Machine (SVM) exhibited the most consistent performance in FGR prediction. SHAP showed that the top contributors to identify FGR were Abdominal Circumference (AC), NF, Uterine RI (Ut RI), and Uterine PI (Ut PI). ALE showed that the cutoff values of Ut RI, Ut PI, and AC in differentiating FGR from normal were comparable with clinical guidelines (Errors between model and clinical; Ut RI: 15%, Ut PI: 8%, and AC: 11%). The cutoff value for NF to differentiate between healthy and FGR is 5.4 mm, where low NF may indicate FGR. The SVM model is the most stable in FGR prediction. ALE can be a potential tool to identify a cutoff value for novel parameters to differentiate between healthy and FGR.

Metabolomics analysis of placental tissue obtained from patients with fetal growth restriction

AIM

The aim of this study was to determine whether there was a difference in placental metabolite profiles between patients with fetal growth restriction (FGR) and healthy controls.

METHODS

The study included 10 patients with FGR diagnosis with 14 healthy controls with both matched maternal age and body mass index. ¹ H HR-MAS NMR spectroscopy data obtained from placental tissue samples of patients with FGR and healthy control group were analyzed with bioinformatics methods. The obtained results of metabolite levels were further validated with the internal standard (IS) quantification method.

RESULTS

Principal component analysis (PCA) and the partial least squares discriminant analysis (PLS-DA) score plots obtained with the multivariate statistical analysis of preprocessed spectral data shows a separation between the samples from patients with FGR and healthy controls. Bioinformatics analysis results suggest that the placental levels of lactate, glutamine, glycerophosphocholine, phosphocholine, taurine, and myoinositol are increased in patients with FGR compared to the healthy controls.

CONCLUSIONS

Placental metabolic dysfunctions are a common occurrence in FGR.

A Comparison of Mother’s Milk and the Neonatal Urine Metabolome: A Unique Fingerprinting for Different Nutritional Phenotypes

The ability of metabolomics to provide a snapshot of an individual’s metabolic state makes it a very useful technique in neonatology for investigating the complex relationship between nutrition and the state of health of the newborn. Through an ¹H-NMR metabolomics analysis, we aimed to investigate the metabolic profile of newborns by analyzing both urine and milk samples in relation to the birth weight of neonates classified as AGA (adequate for the gestational age, n = 51), IUGR (intrauterine growth restriction, n = 14), and LGA (large for gestational age, n = 15). Samples were collected at 7 ± 2 days after delivery. Of these infants, 42 were exclusively breastfed, while 38 received mixed feeding with a variable amount of commercial infant formula (less than 40%) in addition to breast milk. We observed a urinary spectral pattern for oligosaccharides very close to that of the corresponding mother’s milk in the case of exclusively breastfed infants, thus mirroring the maternal phenotype. The absence of this good match between the infant urine and human milk spectra in the case of mixed-fed infants could be reasonably ascribed to the use of a variable amount of commercial infant formulas (under 40%) added to breast milk. Furthermore, our findings did not evidence any significant differences in the spectral profiles in terms of the neonatal customize centile, i.e., AGA (adequate for gestational age), LGA (large for gestational age), or IGUR (intrauterine growth restriction). It is reasonable to assume that maternal human milk oligosaccharide (HMO) production is not or is only minimally influenced by the fetal growth conditions for unknown reasons. This hypothesis may be supported by our metabolomics-based results, confirming once again the importance of this approach in the neonatal field.

Artificial intelligence and placental DNA methylation: newborn prediction and molecular mechanisms of autism in preterm children

BACKGROUND

Autism Spectrum Disorder (ASD) represents a heterogeneous group of disorders with a complex genetic and epigenomic etiology. DNA methylation is the most extensively studied epigenomic mechanism and correlates with altered gene expression. Artificial intelligence (AI) is a powerful tool for group segregation and for handling the large volume of data generated in omics experiments.

METHODS

We performed genome-wide methylation analysis for differential methylation of cytosine nucleotide (CpG) was performed in 20 postpartum placental tissue samples from preterm births. Ten newborns went on to develop autism (Autistic Disorder subtype) and there were 10 unaffected controls. AI including Deep Learning (AI-DL) platforms were used to identify and rank cytosine methylation markers for ASD detection. Ingenuity Pathway Analysis (IPA) to identify genes and molecular pathways that were dysregulated in autism.

RESULTS

We identified 4870 CpG loci comprising 2868 genes that were significantly differentially methylated in ASD compared to controls. Of these 431 CpGs met the stringent EWAS threshold (p-value <5 × 10^-8) along with ≥10% methylation difference between CpGs in cases and controls. DL accurately predicted autism with an AUC (95% CI) of 1.00 (1-1) and sensitivity and specificity of 100% using a combination of 5 CpGs [cg13858611 (NRN1), cg09228833 (ZNF217), cg06179765 (GPNMB), cg08814105 (NKX2-5), cg27092191 (ZNF267)] CpG markers. IPA identified five prenatally dysregulated molecular pathways linked to ASD.

CONCLUSIONS

The present study provides substantial evidence that epigenetic differences in placental tissue are associated with autism development and raises the prospect of early and accurate detection of the disorder.

Placental DNA methylation changes and the early prediction of autism in full-term newborns

Autism spectrum disorder (ASD) is associated with abnormal brain development during fetal life. Overall, increasing evidence indicates an important role of epigenetic dysfunction in ASD. The placenta is critical to and produces neurotransmitters that regulate fetal brain development. We hypothesized that placental DNA methylation changes are a feature of the fetal development of the autistic brain and importantly could help to elucidate the early pathogenesis and prediction of these disorders. Genome-wide methylation using placental tissue from the full-term autistic disorder subtype was performed using the Illumina 450K array. The study consisted of 14 cases and 10 control subjects. Significantly epigenetically altered CpG loci (FDR p-value <0.05) in autism were identified. Ingenuity Pathway Analysis (IPA) was further used to identify molecular pathways that were over-represented (epigenetically dysregulated) in autism. Six Artificial Intelligence (AI) algorithms including Deep Learning (DL) to determine the predictive accuracy of CpG markers for autism detection. We identified 9655 CpGs differentially methylated in autism. Among them, 2802 CpGs were inter- or non-genic and 6853 intragenic. The latter involved 4129 genes. AI analysis of differentially methylated loci appeared highly accurate for autism detection. DL yielded an AUC (95% CI) of 1.00 (1.00-1.00) for autism detection using intra- or intergenic markers by themselves or combined. The biological functional enrichment showed, four significant functions that were affected in autism: quantity of synapse, microtubule dynamics, neuritogenesis, and abnormal morphology of neurons. In this preliminary study, significant placental DNA methylation changes. AI had high accuracy for the prediction of subsequent autism development in newborns. Finally, biologically functional relevant gene pathways were identified that may play a significant role in early fetal neurodevelopmental influences on later cognition and social behavior.

Gestational age-dependent development of the neonatal metabolome

BACKGROUND

Prematurity is a severe pathophysiological condition, however, little is known about the gestational age-dependent development of the neonatal metabolome.

METHODS

Using an untargeted liquid chromatography-tandem mass spectrometry metabolomics protocol, we measured over 9000 metabolites in 298 neonatal residual heel prick dried blood spots retrieved from the Danish Neonatal Screening Biobank. By combining multiple state-of-the-art metabolome mining tools, we retrieved chemical structural information at a broad level for over 5000 (60%) metabolites and assessed their relation to gestational age.

RESULTS

A total of 1459 (~16%) metabolites were significantly correlated with gestational age (false discovery rate-adjusted P < 0.05), whereas 83 metabolites explained on average 48% of the variance in gestational age. Using a custom algorithm based on hypergeometric testing, we identified compound classes (617 metabolites) overrepresented with metabolites correlating with gestational age (P < 0.05). Metabolites significantly related to gestational age included bile acids, carnitines, polyamines, amino acid-derived compounds, nucleotides, phosphatidylcholines and dipeptides, as well as treatment-related metabolites, such as antibiotics and caffeine.

CONCLUSIONS

Our findings elucidate the gestational age-dependent development of the neonatal blood metabolome and suggest that the application of metabolomics tools has great potential to reveal novel biochemical underpinnings of disease and improve our understanding of complex pathophysiological mechanisms underlying prematurity-associated disorders.

IMPACT

A large variation in the neonatal dried blood spot metabolome from residual heel pricks stored at the Danish Neonatal Screening Biobank can be explained by gestational age. While previous studies have assessed the relation of selected metabolic markers to gestational age, this study assesses metabolome-wide changes related to prematurity. Using a combination of recently developed metabolome mining tools, we assess the relation of over 9000 metabolic features to gestational age. The ability to assess metabolome-wide changes related to prematurity in neonates could pave the way to finding novel biochemical underpinnings of health complications related to preterm birth.

Artificial intelligence and leukocyte epigenomics: Evaluation and prediction of late-onset Alzheimer's disease

We evaluated the utility of leucocyte epigenomic-biomarkers for Alzheimer’s Disease (AD) detection and elucidates its molecular pathogeneses. Genome-wide DNA methylation analysis was performed using the Infinium MethylationEPIC BeadChip array in 24 late-onset AD (LOAD) and 24 cognitively healthy subjects. Data were analyzed using six Artificial Intelligence (AI) methodologies including Deep Learning (DL) followed by Ingenuity Pathway Analysis (IPA) was used for AD prediction. We identified 152 significantly (FDR p<0.05) differentially methylated intragenic CpGs in 171 distinct genes in AD patients compared to controls. All AI platforms accurately predicted AD with AUCs ≥0.93 using 283,143 intragenic and 244,246 intergenic/extragenic CpGs. DL had an AUC = 0.99 using intragenic CpGs, with both sensitivity and specificity being 97%. High AD prediction was also achieved using intergenic/extragenic CpG sites (DL significance value being AUC = 0.99 with 97% sensitivity and specificity). Epigenetically altered genes included CR1L & CTSV (abnormal morphology of cerebral cortex), S1PR1 (CNS inflammation), and LTB4R (inflammatory response). These genes have been previously linked with AD and dementia. The differentially methylated genes CTSV & PRMT5 (ventricular hypertrophy and dilation) are linked to cardiovascular disease and of interest given the known association between impaired cerebral blood flow, cardiovascular disease, and AD. We report a novel, minimally invasive approach using peripheral blood leucocyte epigenomics, and AI analysis to detect AD and elucidate its pathogenesis.

Identification of Latent Risk Clinical Attributes for Children Born Under IUGR Condition Using Machine Learning Techniques

BACKGROUND AND OBJECTIVE

Intrauterine Growth Restriction (IUGR) is a condition in which a fetus does not grow to the expected weight during pregnancy. There are several well documented causes in the literature for this issue, such as maternal disorder, and genetic influences. Nevertheless, besides the risk during pregnancy and labour periods, in a long term perspective, the impact of IUGR condition during the child development is an area of research itself. The main objective of this work is to propose a machine learning solution to identify the most significant features of importance based on physiological, clinical or socioeconomic factors correlated with previous IUGR condition after 10 years of birth.

METHODS

In this work, 41 IUGR (18 male) and 34 Non-IUGR (22 male) children were followed up 9 years after the birth, in average (9.1786 ± 0.6784 years old). A group of machine learning algorithms is proposed to classify children previously identified as born under IUGR condition based on 24-hours monitoring of ECG (Holter) and blood pressure (ABPM), and other clinical and socioeconomic attributes. In additional, an algorithm of relevance determination based on the classifier is also proposed, to determine the level of importance of the considered features.

RESULTS

The proposed classification solution achieved accuracy up to 94.73%, and better performance than seven state-of-the-art machine learning algorithms. Also, relevant latent factors related to HRV and BP monitoring are proposed, such as: day-time heart rate (day-time HR), day-night systolic blood pressure (day-night SBP), 24-hour standard deviation (SD) of SBP, dropped, morning cortisol creatinine, 24-hour mean of SDs of all NN intervals for each 5 minutes segment (24-hour SDNNi), among others.

CONCLUSION

With outstanding accuracy of our proposed solutions, the classification system and the indication of relevant attributes may support medical teams on the clinical monitoring of IUGR children during their childhood development.

Classification of intrauterine growth restriction at 34-38 weeks gestation with machine learning models

Objective:

Intrauterine growth restriction (IUGR) is one of the most common causes of stillbirths. The objective of this study is to develop a machine learning model that will be able to accurately and consistently predict whether the estimated fetal weight (EFW) will be below the 10th percentile at 34+0–37 + 6 week’s gestation stage, by using data collected at 20 + 0 to 23 + 6 weeks gestation.

Methods:

Recruitment for the prospective Safe Passage Study (SPS) was done over 7.5 years (2007–2015). An essential part of the fetal assessment was the non-invasive transabdominal recording of the maternal and fetal electrocardiograms as well as the performance of an ultrasound examination for Doppler flow velocity waveforms and fetal biometry at 20 + 0 to 23 + 6 and 34 + 0 to 37 + 6 week’s gestation. Several predictive models were constructed, using supervised learning techniques, and evaluated using the Stochastic Gradient Descent, k-Nearest Neighbours, Logistic Regression and Random Forest methods.

Results:

The final model performed exceptionally well across all evaluation metrics, particularly so for the Stochastic Gradient Descent method: achieving a 93% average for Classification Accuracy, Recall, Precision and F1-Score when random sampling is used and 91% for cross-validation (both methods using a 95% confidence interval). Furthermore, the model identifies the Umbilical Artery Pulsality Index to be the strongest identifier for the prediction of IUGR – matching the literature. Three of the four evaluation methods used achieved above 90% for both True Negative and True Positive results. The ROC Analysis showed a very strong True Positive rate (y-axis) for both target attribute outcomes – AUC value of 0.771.

Conclusions:

The model performs exceptionally well in all evaluation metrics, showing robustness and flexibility as a predictive model for the binary target attribute of IUGR. This accuracy is likely due to the value added by the pre-processed features regarding the fetal gained beats and accelerations, something otherwise absent from previous multi-disciplinary studies. The success of the proposed predictive model allows the pursuit of further birth-related anomalies, providing a foundation for more complex models and lesser-researched subject matter. The data available for this model was a vital part of its success but might also become a limiting factor for further analyses. Further development of similar models could result in better classification performance even with little data available.

Prediction of the development of islet autoantibodies through integration of environmental, genetic, and metabolic markers

BACKGROUND

The Environmental Determinants of the Diabetes in the Young (TEDDY) study has prospectively followed, from birth, children at increased genetic risk of type 1 diabetes. TEDDY has collected heterogenous data longitudinally to gain insights into the environmental and biological mechanisms driving the progression to persistent islet autoantibodies.

METHODS

We developed a machine learning model to predict imminent transition to the development of persistent islet autoantibodies based on time-varying metabolomics data integrated with time-invariant risk factors (eg, gestational age). The machine learning was initiated with 221 potential features (85 genetic, 5 environmental, 131 metabolomic) and an ensemble-based feature evaluation was utilized to identify a small set of predictive features that can be interrogated to better understand the pathogenesis leading up to persistent islet autoimmunity.

RESULTS

The final integrative machine learning model included 42 disparate features, returning a cross-validated receiver operating characteristic area under the curve (AUC) of 0.74 and an AUC of ~0.65 on an independent validation dataset. The model identified a principal set of 20 time-invariant markers, including 18 genetic markers (16 single nucleotide polymorphisms [SNPs] and two HLA-DR genotypes) and two demographic markers (gestational age and exposure to a prebiotic formula). Integration with the metabolome identified 22 supplemental metabolites and lipids, including adipic acid and ceramide d42:0, that predicted development of islet autoantibodies.

CONCLUSIONS

The majority (86%) of metabolites that predicted development of islet autoantibodies belonged to three pathways: lipid oxidation, phospholipase A2 signaling, and pentose phosphate, suggesting that these metabolic processes may play a role in triggering islet autoimmunity.

Glycerophospholipid and detoxification pathways associated with small for gestation age pathophysiology: discovery metabolomics analysis in the SCOPE cohort

INTRODUCTION

Small for gestational age (SGA) may be associated with neonatal morbidity and mortality. Our understanding of the molecular pathways implicated is poor.

OBJECTIVES

Our aim was to determine the metabolic pathways involved in the pathophysiology of SGA and examine their variation between maternal biofluid samples.

METHODS

Plasma (Cork) and urine (Cork, Auckland) samples were collected at 20 weeks' gestation from nulliparous low-risk pregnant women participating in the SCOPE study. Women who delivered an SGA infant (birthweight < 10th percentile) were matched to controls (uncomplicated pregnancies). Metabolomics (urine) and lipidomics (plasma) analyses were performed using ultra performance liquid chromatography-mass spectrometry. Features were ranked based on FDR adjusted p-values from empirical Bayes analysis, and significant features putatively identified.

RESULTS

Lipidomics plasma analysis revealed that 22 out of the 33 significantly altered lipids annotated were glycerophospholipids; all were detected in higher levels in SGA. Metabolomic analysis identified reduced expression of metabolites associated with detoxification (D-Glucuronic acid, Estriol-16-glucuronide), nutrient absorption and transport (Sulfolithocholic acid) pathways.

CONCLUSIONS

This study suggests higher levels of glycerophospholipids, and lower levels of specific urine metabolites are implicated in the pathophysiology of SGA. Further research is needed to confirm these findings in independent samples.

Artificial intelligence enabled applications in kidney disease

Artificial intelligence (AI) is considered as the next natural progression of traditional statistical techniques. Advances in analytical methods and infrastructure enable AI to be applied in health care. While AI applications are relatively common in fields like ophthalmology and cardiology, its use is scarcely reported in nephrology. We present the current status of AI in research toward kidney disease and discuss future pathways for AI. The clinical applications of AI in progression to end-stage kidney disease and dialysis can be broadly subdivided into three main topics: (a) predicting events in the future such as mortality and hospitalization; (b) providing treatment and decision aids such as automating drug prescription; and (c) identifying patterns such as phenotypical clusters and arteriovenous fistula aneurysm. At present, the use of prediction models in treating patients with kidney disease is still in its infancy and further evidence is needed to identify its relative value. Policies and regulations need to be addressed before implementing AI solutions at the point of care in clinics. AI is not anticipated to replace the nephrologists' medical decision-making, but instead assist them in providing optimal personalized care for their patients.

Accelerations of the Fetal Heart Rate in the Screening for Fetal Growth Restriction at 34-38 Week's Gestation

OBJECTIVES

To use machine learning to determine what information on Doppler velocimetry and maternal and fetal heart rates, collected at 20-24 weeks gestation, correlates best with fetal growth restriction according to the estimated fetal weight at 34-38 weeks.

STUDY DESIGN

Data of 4496 pregnant women, collected prospectively for the Safe Passage Study, from August 2007 to August 2016, were used for the present analysis. Doppler flow velocity of the uterine, umbilical, and middle cerebral arteries and transabdominally recorded maternal and fetal ECGs were collected at 20-24 weeks gestation and fetal biometry collected at 34-38 weeks from which the estimated fetal weight was calculated. Fetal growth restriction was defined as an estimated fetal weight below the 10th centile. Accelerations and decelerations of the fetal and maternal heart rates were quantified as gained or lost beats per hour of recording respectively. Machine learning with receiver operative characteristic curves were then used to determine which model gives the best performance.

RESULTS

The final model performed exceptionally well across all evaluation metrics, particularly so for the Stochastic Gradient Descent method: achieving a 93% average for Classification Accuracy, Recall, Precision and F1-Score to identify the fetus with an estimated weight below the 10th percentile at 34-38 weeks. Ranking determined that the most important standard feature was the umbilical artery pulsatility index. However, the excellent overall accuracy is likely due to the value added by the pre-processed features regarding fetal gained beats and accelerations.

CONCLUSION

Fetal movements, as characterized by gained beats as early as 20-24 weeks gestation, contribute to the value of the flow velocimetry of the umbilical artery at 34-38 weeks in identifying the growth restricted fetus.

A Community-Based Study Identifying Metabolic Biomarkers of Mild Cognitive Impairment and Alzheimer's Disease Using Artificial Intelligence and Machine Learning

BACKGROUND

Currently, there is no objective, clinically available tool for the accurate diagnosis of Alzheimer's disease (AD). There is a pressing need for a novel, minimally invasive, cost friendly, and easily accessible tool to diagnose AD, assess disease severity, and prognosticate course. Metabolomics is a promising tool for discovery of new, biologically, and clinically relevant biomarkers for AD detection and classification.

OBJECTIVE

Utilizing artificial intelligence and machine learning, we aim to assess whether a panel of metabolites as detected in plasma can be used as an objective and clinically feasible tool for the diagnosis of mild cognitive impairment (MCI) and AD.

METHODS

Using a community-based sample cohort acquired from different sites across the US, we adopted an approach combining Proton Nuclear Magnetic Resonance Spectroscopy (1H NMR), Liquid Chromatography coupled with Mass Spectrometry (LC-MS) and various machine learning statistical approaches to identify a biomarker panel capable of identifying those patients with AD and MCI from healthy controls.

RESULTS

Of the 212 measured metabolites, 5 were identified as optimal to discriminate between controls, and individuals with MCI or AD. Our models performed with AUC values in the range of 0.72-0.76, with the sensitivity and specificity values ranging from 0.75-0.85 and 0.69-0.81, respectively. Univariate and pathway analysis identified lipid metabolism as the most perturbed biochemical pathway in MCI and AD.

CONCLUSION

A comprehensive method of acquiring metabolomics data, coupled with machine learning techniques, has identified a strong panel of diagnostic biomarkers capable of identifying individuals with MCI and AD. Further, our data confirm what other groups have reported, that lipid metabolism is significantly perturbed in those individuals suffering with dementia. This work may provide additional insight into AD pathogenesis and encourage more in-depth analysis of the AD lipidome.

[Analysis of metabolic pathways in intrauterine growth restriction]

Objective was to analyze metabolic pathways based on a study of the metabolomic profile of pregnant women with intrauterine growth restriction. The metabolic profile of pregnant women with fetal growth restriction has been analyzed using liquid chromatography-mass spectrometry. At the second stage pathways were identified using SMPDB and MetaboAnalyst databases to clarify the relationship between metabolites. Biological networks allow to determine the effect of proteins on the metabolic pathways involved in pathogenesis of IUGR and determine the epigenetic mechanisms of its formation.

Predicting risk of low birth weight offspring from maternal features and blood polycyclic aromatic hydrocarbon concentration

Prenatal exposure to organic pollutants increases the risk of low birth weight (LBW) offspring. Women involved in the plucking of tea leaves can be exposed to polycyclic aromatic hydrocarbons (PAHs) during pregnancy through inhalation and diet. Therefore, the aim of the study was to investigate the association of maternal socio-demographic features and blood PAH concentration with LBW; also to develop a model for predicting LBW risk. The study was performed by recruiting 55 women who delivered LBW and 120 women with NBW (normal birth weight) babies from Assam Medical College. The placental tissue, maternal and cord blood samples were collected. A total of sixteen PAHs and cotinine were analysed by HPLC and GC-MS. Association of PAH concentration with weight was determined using correlation and multiple logistic regression analyses. Predictive model was developed using SVM^light and Weka software. Maternal features such as age, education, food habits, occupation, etc. were found to be associated with LBW deliveries (p-value<0.05). Overall, 9 PAHs and cotinine were detected in the samples. A multiple logistic regression depicted an increased likelihood of LBW by exposure to PAHs (pyrene, di-benzo (a,h) anthracene, fluorene and fluoranthene) and cotinine. Models based on the features and PAHs/ cotinine predicted LBW offspring with 84.35% sensitivity and 74% specificity. LBW prediction models are available at http://dev.icmr.org.in/plbw/ webserver. With machine learning gaining more importance in medical science; our webserver could be instrumental for researchers and clinicians to predict the state of the fetus.

How could metabolomics change pediatric health?

In the last years, 'omics' technologies, and especially metabolomics, emerged as expanding scientific disciplines and promising technologies in the characterization of several pathophysiological processes.In detail, metabolomics, able to detect in a dynamic way the whole set of molecules of low molecular weight in cells, tissues, organs, and biological fluids, can provide a detailed phenotypic portray, representing a metabolic "snapshot."Thanks to its numerous strength points, metabolomics could become a fundamental tool in human health, allowing the exact evaluation of individual metabolic responses to pathophysiological stimuli including drugs, environmental changes, lifestyle, a great number of diseases and other epigenetics factors.Moreover, if current metabolomics data will be confirmed on larger samples, such technology could become useful in the early diagnosis of diseases, maybe even before the clinical onset, allowing a clinical monitoring of disease progression and helping in performing the best therapeutic approach, potentially predicting the therapy response and avoiding overtreatments. Moreover, the application of metabolomics in nutrition could provide significant information on the best nutrition regimen, optimal infantile growth and even in the characterization and improvement of commercial products' composition.These are only some of the fields in which metabolomics was applied, in the perspective of a precision-based, personalized care of human health.In this review, we discuss the available literature on such topic and provide some evidence regarding clinical application of metabolomics in heart diseases, auditory disturbance, nephrouropathies, adult and pediatric cancer, obstetrics, perinatal conditions like asphyxia, neonatal nutrition, neonatal sepsis and even some neuropsychiatric disorders, including autism.Our research group has been interested in metabolomics since several years, performing a wide spectrum of experimental and clinical studies, including the first metabolomics analysis of human breast milk. In the future, it is reasonable to predict that the current knowledge could be applied in daily clinical practice, and that sensible metabolomics biomarkers could be easily detected through cheap and accurate sticks, evaluating biofluids at the patient's bed, improving diagnosis, management and prognosis of sick patients and allowing a personalized medicine. A dream? May be I am a dreamer, but I am not the only one.

Orthogonal Method for Double-Bond Placement via Ozone-Induced Dissociation Mass Spectrometry (OzID-MS)

Most often, the structures of secondary metabolites are solved using a suite of NMR techniques. However, there are times when it can be challenging to position double bonds, particularly those that are fully substituted or when there are multiple double bonds in similar chemical environments. Ozone-induced dissociation mass spectrometry (OzID-MS) serves as an orthogonal structure elucidation tool, using predictable fragmentation patterns that are generated after ozonolysis across a carbon-carbon double bond. This technique is finding growing use in the lipidomics community, suggestive of its potential value for secondary metabolites. This methodology was evaluated by confirming the double-bond positions in five fungal secondary metabolites, specifically, ent-sartorypyrone E (1), sartorypyrone A (2), sorbicillin (3), trichodermic acid A (4), and AA03390 (5). This demonstrated its potential with a variety of chemotypes, ranging from polyketides to terpenoids and including those in both conjugated and nonconjugated polyenes. In addition, the potential of using this methodology in the context of a mixture was piloted by studying Aspergillus fischeri, first examining a traditional extract and then sampling a live fungal culture in situ. While the intensity of signals varied from pure compound to extract to in situ, the utility of the technique was preserved.

The plasma metabolome of women in early pregnancy differs from that of non-pregnant women

BACKGROUND

In comparison to the non-pregnant state, the first trimester of pregnancy is characterized by systemic adaptation of the mother. The extent to which these adaptive processes are reflected in the maternal blood metabolome is not well characterized.

OBJECTIVE

To determine the differences between the plasma metabolome of non-pregnant and pregnant women before 16 weeks gestation.

STUDY DESIGN

This study included plasma samples from 21 non-pregnant women and 50 women with a normal pregnancy (8-16 weeks of gestation). Combined measurements by ultrahigh performance liquid chromatography/tandem mass spectrometry and by gas chromatography/mass spectrometry generated molecular abundance measurements for each sample. Molecular species detected in at least 10 samples were included in the analysis. Differential abundance was inferred based on false discovery adjusted p-values (FDR) from Mann-Whitney-Wilcoxon U tests <0.1 and a minimum median abundance ratio (fold change) of 1.5. Alternatively, metabolic data were quantile normalized to remove sample-to-sample differences in the overall metabolite abundance (adjusted analysis).

RESULTS

Overall, 637 small molecules met the inclusion criteria and were tested for association with pregnancy; 44% (281/637) of small molecules had significantly different abundance, of which 81% (229/281) were less abundant in pregnant than in non-pregnant women. Eight percent (14/169) of the metabolites that remained significant in the adjusted analysis also changed as a function of gestational age. A pathway analysis revealed enrichment in steroid metabolites related to sex hormones, caffeine metabolites, lysolipids, dipeptides, and polypeptide bradykinin derivatives (all, FDR < 0.1).

CONCLUSIONS

This high-throughput mass spectrometry study identified: 1) differences between pregnant vs. non-pregnant women in the abundance of 44% of the profiled plasma metabolites, including known and novel molecules and pathways; and 2) specific metabolites that changed with gestational age.

Intrauterine Growth Restriction: New Insight from the Metabolomic Approach

Recognizing intrauterine growth restriction (IUGR) is a matter of great concern because this condition can significantly affect the newborn's short- and long-term health. Ever since the first suggestion of the "thrifty phenotype hypothesis" in the last decade of the 20th century, a number of studies have confirmed the association between low birth weight and cardiometabolic syndrome later in life. During intrauterine life, the growth-restricted fetus makes a number of hemodynamic, metabolic, and hormonal adjustments to cope with the adverse uterine environment, and these changes may become permanent and irreversible. Despite advances in our knowledge of IUGR newborns, biomarkers capable of identifying this condition early on, and stratifying its severity both pre- and postnatally, are still lacking. We are also still unsure about these babies' trajectory of postnatal growth and their specific nutritional requirements with a view to preventing, or at least limiting, long-term complications. In this setting, untargeted metabolomics-a relatively new field of '-omics' research-can be a good way to investigate the metabolic perturbations typically associated with IUGR. The aim of this narrative review is to provide a general overview of the pathophysiological and clinical aspects of IUGR, focusing on evidence emerging from metabolomic studies. Though still only preliminary, the reports emerging so far suggest an "early" pattern of glucose intolerance, insulin resistance, catabolite accumulation, and altered amino acid metabolism in IUGR neonates. Further, larger studies are needed to confirm these results and judge their applicability to clinical practice.