Integración de datos multi-ómicos mediante machine learning para entender los mecanismos moleculares de la resistencia a la insulina en niños con obesidad durante el desarrollo puberal
Year: 2026
Type: dissertation
Study design: Observational Study
Abstract: Childhood obesity is a major global health challenge and a key driver of early metabolic dysfunction. Among its cardiometabolic complications, insulin resistance (IR) emerges during paediatric life and frequently worsens across puberty, a developmental window characterised by profound hormonal, metabolic and adipose tissue remodelling. While transient physiological IR typically resolves in normal-weight youth, children with obesity often develop persistent metabolic impairment, increasing their long-term risk of type 2 diabetes and cardiovascular disease. The biological mechanisms underlying this divergence remain incompletely understood. IR during puberty arises from complex and dynamic interactions across multiple omics layers, including genetic susceptibility and epigenetic regulation. Conventional cross-sectional approaches fail to capture the temporal molecular trajectories that shape disease progression. Longitudinal multi-omics strategies, combined with bioinformatic frameworks, are therefore required to disentangle mechanistic pathways and refine early cardiometabolic risk assessment. This Doctoral Thesis aimed to elucidate the molecular architecture of obesity-associated IR across the pubertal transition and to develop integrative multi-omics and machine learning (ML) frameworks for early risk prediction in children with obesity. First, the problem of methodological heterogeneity in paediatric cardiometabolic assessment was addressed through the development of ObMetrics, a web-based application enabling harmonised computation of metabolic syndrome components and age- and sex-specific z-scores. Second, the first longitudinal epigenome-wide association study of obesity-associated IR across puberty in Spanish children was conducted, identifying dynamic DNA methylation patterns associated with IR progression, persistence or improvement. These findings position puberty as a critical window of epigenetic remodelling linked to metabolic vulnerability. Third, vasorin was identified as a robust biomarker of paediatric obesity and cardiometabolic dysfunction, supported by cross-omics concordance across DNA methylation, gene expression and circulating protein levels in independent cohorts and tissues, strengthening biological coherence and translational relevance. Finally, an early integrative multi-omics ML framework combining genetic, epigenetic and clinical data demonstrated accurate prediction of pubertal IR from prepubertal molecular profiles, serving as a proof of concept for its future translation into clinical decision support system. The incorporation of explainable artificial intelligence approaches ensured interpretability and transparency, bridging predictive performance with clinical applicability. Together, these findings demonstrate that longitudinal multi-omics integration, supported by bioinformatic pipelines and interpretable ML models, enables a transition from descriptive molecular associations to personalised risk prediction in paediatric obesity. By capturing temporal molecular dynamics and leveraging integrative predictive modelling, this work contributes to advancing precision prevention strategies aimed at mitigating the lifelong cardiometabolic burden associated with childhood obesity.
Author: Torres-Martos, Álvaro
Language: English
Subfield: Endocrinology, Diabetes and Metabolism
Field: Medicine
Domain: Health Sciences
Open Access status: green
Awards: G1091926076, G1622782223, G1662035387, Preventing lifetime obesity by early risk-factor identification, prognosis and intervention, OMICAS E INTELINGENCIA ARTIFICAL. HERRAMIENTAS PARA ENTENDER LOS MECANISMOS MOLECULARES DE LA RESISTENCIA A INSULINA EN NIÑOS OBESOS DURANTE LA PUBERTAD., G236855531, G2620658983, Hacia una Medicina de Precisión: Análisis del Exposoma y Omicas usando Inteligencia Artificial Explicable para estudiar su impacto durante la vida en la Obesidad, InsulinoResistencia y salud Metabólica. EXOMAIR, Hacia una Medicina de Precisión: Análisis del Exposoma y Omicas usando Inteligencia Artificial Explicable para estudiar su impacto durante la vida en la Obesidad, Insulino Resistencia y salud metabólica EXOMAIR, G3034612882, G3480869486, G3588086252, G3989941764, G3992112793, G4006774809, OMICAS E INTELINGENCIA ARTIFICAL. HERRAMIENTAS PARA ENTENDER LOS MECANISMOS MOLECULARES DE LA RESISTENCA A INSULINA EN NIÑOS OBESOS DURANTE LA PUBERTAD., Hacia una Medicina de Precisión: Análisis del Exposoma y Ómicas usando Inteligencia Artificial eXplicable para estudiar su impacto durante la vida en la Obesidad, InsulinoRresistencia y salud Metabólica. EXOMAIR, Hacia una Medicina de Precisión: Análisis del Exposoma y Omicas usando Inteligencia Artificial Explicable para estudiar su impacto durante la vida en la Obesidad, Insulino Resistencia y salud metabólica. EXOMAIR, G4521193091, G4526231891, G5034114885, G5080475149, Hacia la Medicina de Precisión: Análisis del Exposoma y Omicas usando Inteligencia Artificial Explicable para estudiar su impacto durante la vida en la Obesidad, Insulino Resistencia y salud metabólica. EXOMAIR, G5467699721, G6434112895, G7084143925, G7728232459, G8177970503, G8211200720, Enfoque Machine Learning y Big Data Multi-Ómico: Medicina Personalizada
OpenAlex ID: W7213381054