Alex Sánchez-Pla

The Mediterranean diet (MD) increased pyruvate, acetoacetate and TCA intermediate levels, and restored BCAA metabolism. Some of these changes related to metabolic flexibility were associated with improved cognition in older adults with MCI.

High-dimensional omic datasets present major challenges for machine learning due to their sparse biological signal, strong feature heterogeneity, and high dimensionality. In this work, we propose PLAT (Parallel Latent Attention Transformer), a neural architecture for high-dimensional tabular transcriptomic data. The model projects input gene expression features into multiple parallel latent representations, each processed independently through self-attention to capture complementary feature interactions while maintaining moderate model complexity. The proposed architecture was evaluated using both controlled Negative Binomial simulations designed to reproduce RNA-seq overdispersion and the TCGA-BRCA breast cancer dataset comprising 499 patients and 4376 gene expression variables for ER+/ER− classification. Comparative analyses against a baseline multilayer perceptron and a lightweight FT-Transformer showed that PLAT achieves competitive predictive performance while maintaining a comparable number of trainable parameters. Simulation experiments further indicate that its main advantage is concentrated in specific high-dimensional settings with an intermediate proportion of informative features. To assess model interpretability, we additionally performed a SHAP-based analysis of the baseline MLP and compared it with the attention-derived gene rankings. Although both models identified largely different sets of predictive genes, functional enrichment analyses consistently highlighted biological processes and disease pathways associated with breast cancer, supporting the biological relevance of the learned latent representations. These results suggest that PLAT provides an effective and interpretable framework for high-dimensional transcriptomic classification.

Background: The impact of chronic medication on COVID-19 outcomes has been a topic of ongoing debate since the onset of the pandemic. Investigating how specific long-term treatments influence infection severity and prognosis is essential for optimising patient management and care. Aim: This study aimed to investigate the association between chronic medication and COVID-19 outcomes, using machine learning to identify key medication-related factors. Method: We analysed 137,835 COVID-19 patients in Catalonia (February–September 2020) using eXtreme Gradient Boosting to predict hospitalisation, ICU admission, and mortality. This was complemented by univariate logistic regression analyses and a sensitivity analysis focusing on diabetes, hypertension, and lipid disorders. Results: Participants had a mean age of 53 (SD 20) years, with 57% female. The best model predicted mortality risk in 18 to 65-year-olds (AUCROC 0.89, CI 0.85–0.92). Key features identified included the number of prescribed drugs, systemic corticoids, 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase, and hypertension drugs. A sensitivity analysis identified that hypertensive participants over 65 taking angiotensin-converting enzyme (ACE) inhibitors or angiotensin II receptor blockers (ARBs) had lower mortality risk (OR 0.78 CI 0.68–0.92) compared to those on other antihypertensive medication (OR 0.8 CI 0.68–0.95). Treatment with inhibitors of dipeptidyl peptidase 4 was associated to higher mortality in participants aged 18–65, while metformin showed a protective effect in those over 65 (OR 0.79, 95% CI 0.68–0.92). Conclusion: Machine learning models effectively distinguished COVID-19 outcomes. Patients under ACEi or ARBs or biguanides should continue their prescribed medications, which may offer protection over alternative treatments.

High-throughput technologies produce large-scale omics datasets, and their integration facilitates biomarker discovery and predictive modeling. However, challenges such as data heterogeneity, high dimensionality, and block-wise missing data complicate the analysis. To address these issues, optimization techniques, including regularization and constraint-based approaches, have been already employed for regression and binary classification problems. Building on these methods, we extended this framework to support multi-class classification. Indeed, applied to a multi-class classification task for breast cancer subtypes, our model achieves accuracy between 73% and 81% under various block-wise missing data scenarios. Additionally, we assess its performance on a regression problem using the exposome dataset, integrating a larger number of omics datasets. Across different missing data scenarios, our model demonstrates a strong correlation (75%) between true and predicted responses. Furthermore, we have updated the bwm R package, which previously supported binary and continuous response types, to also include multi-class response types.

Background: The antibody response to SARS-CoV-2 does not follow the immunoglobulin isotype pattern expected in a primary response and is inconsistent with the current interpretation of COVID-19 immunopathology as the result of a primary infection. To better understand the immune response to SARS-CoV-2, it is essential to determine whether it is primary or secondary (or recall). The analysis of highly granular immunological variable trajectories of a homogeneous cohort of patients receiving standardised medical care should discern between primary and secondary responses. Methods: This is a prospective cohort study of 191 SARS-CoV-2 infection cases and 44 healthy controls from the second wave of COVID-19 in the Barcelona area. The study stratified patients by severity and analysed the trajectories of SARS-CoV-2 antibodies and multiple immune variables for features associated with primary and recall immune responses. Findings: Isotype-specific antibody trajectories to SARS-CoV-2 proteins revealed a pattern of recall response in 94·2% of cases. In these cases, the detailed trajectories of plasmablasts, B cells, cTfh high-resolution subsets, and cytokines were consistent with a secondary response. The transcriptomic data indicated that this cohort is strictly comparable to contemporary cohorts. Conclusions: In most cases, the immune response to SARS-CoV-2 is a recall response. This opens the possibility that most COVID-19 cases are subjected to immune imprinting by endemic coronavirus, which, in turn, can contribute to severity by interfering with the immune response to SARS-CoV-2 and by antibody-dependent enhancement. Considering the immune responses to SARS-CoV-2 secondary provides a better perspective to interpret COVID-19 pathology. Funding: Grants COV20/00416, Cov20/00654, and COV20/00388 from Instituto de Salud Carlos III (ISCIII), Madrid, Spain, co-financed by the European Regional Development Fund (ERDF).

Biomedical data are generated and collected from various sources, including medical imaging, laboratory tests and genome sequencing. Sharing these data for research can help address unmet health needs, contribute to scientific breakthroughs, accelerate the development of more effective treatments and inform public health policy. Due to the potential sensitivity of such data, however, privacy concerns have led to policies that restrict data sharing. In addition, sharing sensitive data requires a secure and robust infrastructure with appropriate storage solutions. Here, we examine and compare the centralized and federated data sharing models through the prism of five large-scale and real-world use cases of strategic significance within the European data sharing landscape: the French Health Data Hub, the BBMRI-ERIC Colorectal Cancer Cohort, the federated European Genome-phenome Archive, the Observational Medical Outcomes Partnership/OHDSI network and the EBRAINS Medical Informatics Platform. Our analysis indicates that centralized models facilitate data linkage, harmonization and interoperability, while federated models facilitate scaling up and legal compliance, as the data typically reside on the data generator’s premises, allowing for better control of how data are shared. This comparative study thus offers guidance on the selection of the most appropriate sharing strategy for sensitive datasets and provides key insights for informed decision-making in data sharing efforts.

Hippocampal neurogenesis (HN) occurs throughout the life course and is important for memory and mood. Declining with age, HN plays a pivotal role in cognitive decline (CD), dementia, and late-life depression, such that altered HN could represent a neurobiological susceptibility to these conditions. Pertinently, dietary patterns (e.g., Mediterranean diet) and/or individual nutrients (e.g., vitamin D, omega 3) can modify HN, but also modify risk for CD, dementia, and depression. Therefore, the interaction between diet/nutrition and HN may alter risk trajectories for these ageing-related brain conditions. Using a subsample (n = 371) of the Three-City cohort—where older adults provided information on diet and blood biobanking at baseline and were assessed for CD, dementia, and depressive symptomatology across 12 years—we tested for interactions between food consumption, nutrient intake, and nutritional biomarker concentrations and neurogenesis-centred susceptibility status (defined by baseline readouts of hippocampal progenitor cell integrity, cell death, and differentiation) on CD, Alzheimer’s disease (AD), vascular and other dementias (VoD), and depressive symptomatology, using multivariable-adjusted logistic regression models. Increased plasma lycopene concentrations (OR [95% CI] = 1.07 [1.01, 1.14]), higher red meat (OR [95% CI] = 1.10 [1.03, 1.19]), and lower poultry consumption (OR [95% CI] = 0.93 [0.87, 0.99]) were associated with an increased risk for AD in individuals with a neurogenesis-centred susceptibility. Increased vitamin D consumption (OR [95% CI] = 1.05 [1.01, 1.11]) and plasma γ-tocopherol concentrations (OR [95% CI] = 1.08 [1.01, 1.18]) were associated with increased risk for VoD and depressive symptomatology, respectively, but only in susceptible individuals. This research highlights an important role for diet/nutrition in modifying dementia and depression risk in individuals with a neurogenesis-centred susceptibility.

The main goals and challenges for the life science communities in the Open Science framework are to increase reuse and sustainability of data resources, software tools, and workflows, especially in large‐scale data‐driven research and computational analyses. Here, we present key findings, procedures, effective measures and recommendations for generating and establishing sustainable life science resources based on the collaborative, cross‐disciplinary work done within the EOSC‐Life (European Open Science Cloud for Life Sciences) consortium. Bringing together 13 European life science research infrastructures, it has laid the foundation for an open, digital space to support biological and medical research. Using lessons learned from 27 selected projects, we describe the organisational, technical, financial and legal/ethical challenges that represent the main barriers to sustainability in the life sciences. We show how EOSC‐Life provides a model for sustainable data management according to FAIR (findability, accessibility, interoperability, and reusability) principles, including solutions for sensitive‐ and industry‐related resources, by means of cross‐disciplinary training and best practices sharing. Finally, we illustrate how data harmonisation and collaborative work facilitate interoperability of tools, data, solutions and lead to a better understanding of concepts, semantics and functionalities in the life sciences.

Background: Data for Spain from the Rome Foundation Global Epidemiology Study on the disorders of gut–brain interaction (DGBI) were used to assess the national and regional prevalence of all 22 DGBI, the percentage of respondents meeting diagnostic criteria for at least one DGBI, and the impact on burden of disease in our country.Methods: Data were collected through an anonymous, nationwide, and secure Internet survey with multiple built‐in quality‐assurance techniques that included the Rome IV diagnostic questionnaire and an in‐depth supplemental questionnaire.Key Results: The survey was completed by 2072 adult Spanish participants (50.2% female) with a mean age of 45.67 ± 15.44 years with a good representative national distribution. 43.6% (41.5%–45.8%) met diagnostic criteria for at least one DGBI, with 8.2% for any esophageal disorder, 12.1% for any gastroduodenal disorder, 30.1% for any bowel disorder, and 11.5% for any anorectal disorder. Functional constipation was the most prevalent DGBI in Spain (12.8%). We found that proctalgia fugax (9.3%), unspecified bowel disorders (10.8%), and functional dysphagia (5.6%) showed unexplained high rates in our country. DGBI rates were higher for women. Having any DGBI was negatively associated with psychosocial variables (including quality of life, somatization, and concern about digestive problems), and associated with increased healthcare utilization.Conclusions & Inferences: We provide the first comprehensive data on the prevalence and burden of all DGBI in Spain using the Rome IV criteria. The enormous burden of DGBI in Spain highlights the need for specialized training and future research.

Anthocyanins (ACNs) are (poly)phenols associated with reduced cardiometabolic risk. Associations between dietary intake, microbial metabolism, and cardiometabolic health benefits of ACNs have not been fully characterized. Our aims were to study the association between ACN intake, considering its dietary sources, and plasma metabolites, and to relate them with cardiometabolic risk factors in an observational study. A total of 1351 samples from 624 participants (55% female, mean age: 45 ± 12 years old) enrolled in the DCH-NG MAX study were studied using a targeted metabolomic analysis. Twenty-four-hour dietary recalls were used to collect dietary data at baseline, six, and twelve months. ACN content of foods was calculated using Phenol Explorer and foods were categorized into food groups. The median intake of total ACNs was 1.6mg/day. Using mixed graphical models, ACNs from different foods showed specific associations with plasma metabolome biomarkers. Combining these results with censored regression analysis, metabolites associated with ACNs intake were: salsolinol sulfate, 4-methylcatechol sulfate, linoleoyl carnitine, 3,4-dihydroxyphenylacetic acid, and one valerolactone. Salsolinol sulfate and 4-methylcatechol sulfate, both related to the intake of ACNs mainly from berries, were inversely associated with visceral adipose tissue. In conclusion, plasma metabolome biomarkers of dietary ACNs depended on the dietary source and some of them, such as salsolinol sulfate and 4-methylcatechol sulfate may link berry intake with cardiometabolic health benefits.