Multiomics and eXplainable artificial intelligence for decision support in insulin resistance early diagnosis: A pediatric population-based longitudinal study.

Pediatric obesity can drastically heighten the risk of cardiometabolic alterations later in life, with insulin resistance standing as the cornerstone linking adiposity to the increased cardiovascular risk. Puberty has been pointed out as a critical stage after which obesity-associated insulin resistance is more difficult to revert. Timely prediction of insulin resistance in pediatric obesity is therefore vital for mitigating the risk of its associated comorbidities. The construction of effective and robust predictive systems for a complex health outcome like insulin resistance during the early stages of life demands the adoption of longitudinal designs for more causal inferences, and the integration of factors of varying nature involved in its onset. In this work, we propose an eXplainable Artificial Intelligence-based decision support pipeline for early diagnosis of insulin resistance in a longitudinal cohort of 90 children. For that, we leverage multi-omics (genomics and epigenomics) and clinical data from the pre-pubertal stage. Different data layers combinations, pre-processing techniques (missing values, feature selection, class imbalance, etc.), algorithms, training procedures were considered following good practices for Machine Learning. SHapley Additive exPlanations were provided for specialists to understand both the decision-making mechanisms of the system and the impact of the features on each automatic decision, an essential issue in high-risk areas such as this one where system decisions may affect people's lives. The system showed a relevant predictive ability (AUC and G-mean of 0.92). A deep exploration, both at the global and the local level, revealed promising biomarkers of insulin resistance in our population, highlighting classical markers, such as Body Mass Index z-score or leptin/adiponectin ratio, and novel ones such as methylation patterns of relevant genes, such as HDAC4, PTPRN2, MATN2, RASGRF1 and EBF1. Our findings highlight the importance of integrating multi-omics data and following eXplainable Artificial Intelligence trends when building decision support systems.

Omics Data Preprocessing for Machine Learning: A Case Study in Childhood Obesity.

The use of machine learning techniques for the construction of predictive models of disease outcomes (based on omics and other types of molecular data) has gained enormous relevance in the last few years in the biomedical field. Nonetheless, the virtuosity of omics studies and machine learning tools are subject to the proper application of algorithms as well as the appropriate pre-processing and management of input omics and molecular data. Currently, many of the available approaches that use machine learning on omics data for predictive purposes make mistakes in several of the following key steps: experimental design, feature selection, data pre-processing, and algorithm selection. For this reason, we propose the current work as a guideline on how to confront the main challenges inherent to multi-omics human data. As such, a series of best practices and recommendations are also presented for each of the steps defined. In particular, the main particularities of each omics data layer, the most suitable preprocessing approaches for each source, and a compilation of best practices and tips for the study of disease development prediction using machine learning are described. Using examples of real data, we show how to address the key problems mentioned in multi-omics research (e.g., biological heterogeneity, technical noise, high dimensionality, presence of missing values, and class imbalance). Finally, we define the proposals for model improvement based on the results found, which serve as the bases for future work.

Body mass index interacts with a genetic-risk score for depression increasing the risk of the disease in high-susceptibility individuals.

Depression is strongly associated with obesity among other chronic physical diseases. The latest mega- and meta-analysis of genome-wide association studies have identified multiple risk loci robustly associated with depression. In this study, we aimed to investigate whether a genetic-risk score (GRS) combining multiple depression risk single nucleotide polymorphisms (SNPs) might have utility in the prediction of this disorder in individuals with obesity. A total of 30 depression-associated SNPs were included in a GRS to predict the risk of depression in a large case-control sample from the Spanish PredictD-CCRT study, a national multicentre, randomized controlled trial, which included 104 cases of depression and 1546 controls. An unweighted GRS was calculated as a summation of the number of risk alleles for depression and incorporated into several logistic regression models with depression status as the main outcome. Constructed models were trained and evaluated in the whole recruited sample. Non-genetic-risk factors were combined with the GRS in several ways across the five predictive models in order to improve predictive ability. An enrichment functional analysis was finally conducted with the aim of providing a general understanding of the biological pathways mapped by analyzed SNPs. We found that an unweighted GRS based on 30 risk loci was significantly associated with a higher risk of depression. Although the GRS itself explained a small amount of variance of depression, we found a significant improvement in the prediction of depression after including some non-genetic-risk factors into the models. The highest predictive ability for depression was achieved when the model included an interaction term between the GRS and the body mass index (BMI), apart from the inclusion of classical demographic information as marginal terms (AUC = 0.71, 95% CI = [0.65, 0.76]). Functional analyses on the 30 SNPs composing the GRS revealed an over-representation of the mapped genes in signaling pathways involved in processes such as extracellular remodeling, proinflammatory regulatory mechanisms, and circadian rhythm alterations. Although the GRS on its own explained a small amount of variance of depression, a significant novel feature of this study is that including non-genetic-risk factors such as BMI together with a GRS came close to the conventional threshold for clinical utility used in ROC analysis and improves the prediction of depression. In this study, the highest predictive ability was achieved by the model combining the GRS and the BMI under an interaction term. Particularly, BMI was identified as a trigger-like risk factor for depression acting in a concerted way with the GRS component. This is an interesting finding since it suggests the existence of a risk overlap between both diseases, and the need for individual depression genetics-risk evaluation in subjects with obesity. This research has therefore potential clinical implications and set the basis for future research directions in exploring the link between depression and obesity-associated disorders. While it is likely that future genome-wide studies with large samples will detect novel genetic variants associated with depression, it seems clear that a combination of genetics and non-genetic information (such is the case of obesity status and other depression comorbidities) will still be needed for the optimization prediction of depression in high-susceptibility individuals.

Impact of Physical Activity and Exercise on the Epigenome in Skeletal Muscle and Effects on Systemic Metabolism.

Exercise and physical activity induces physiological responses in organisms, and adaptations in skeletal muscle, which is beneficial for maintaining health and preventing and/or treating most chronic diseases. These adaptations are mainly instigated by transcriptional responses that ensue in reaction to each individual exercise, either resistance or endurance. Consequently, changes in key metabolic, regulatory, and myogenic genes in skeletal muscle occur as both an early and late response to exercise, and these epigenetic modifications, which are influenced by environmental and genetic factors, trigger those alterations in the transcriptional responses. DNA methylation and histone modifications are the most significant epigenetic changes described in gene transcription, linked to the skeletal muscle transcriptional response to exercise, and mediating the exercise adaptations. Nevertheless, other alterations in the epigenetics markers, such as epitranscriptomics, modifications mediated by miRNAs, and lactylation as a novel epigenetic modification, are emerging as key events for gene transcription. Here, we provide an overview and update of the impact of exercise on epigenetic modifications, including the well-described DNA methylations and histone modifications, and the emerging modifications in the skeletal muscle. In addition, we describe the effects of exercise on epigenetic markers in other metabolic tissues; also, we provide information about how systemic metabolism or its metabolites influence epigenetic modifications in the skeletal muscle.