Abstract: The hydrogen isotope composition of leaf wax biomarkers (?2Hwax) is a valuable tool for reconstructing continental paleohydrology, since it serves as a proxy for the hydrogen isotope composition of precipitation (?2Hpre). To yield robust palaeohydrological reconstructions using ?2Hwax in marine archives, it is necessary to examine the impacts of regional climate on ?2Hwax and assess the similarity between marine sedimentary ?2Hwax and the source of continental ?2Hwax. Here, we examined an aridity gradient from hyperarid to humid along the Chilean coast. We sampled sediments at the outlets of rivers draining into the Pacific as well as soils within catchments and marine surface sediments adjacent to the outlets of the studied rivers and analyzed the relationship between climatic variables and ?2Hwax values. We found that apparent fractionation between leaf waxes and source water is relatively constant in humid and semiarid regions (average: ?121 ‰). However, it becomes less negative in hyperarid regions (average: ?86 ‰) as a result of evapotranspirative processes affecting soil and leaf water 2H enrichment. We also observed that along strong aridity gradients, the 2H enrichment of ?2Hwax follows a non-linear relationship with water content and water flux variables, driven by strong soil evaporation and plant transpiration. Furthermore, our results indicate that ?2Hwax values in marine surface sediments largely reflect ?2Hwax values from the continent, confirming the robustness of marine ?2Hwax records for paleohydrological reconstructions along the Chilean margin. These findings also highlight the importance of considering the effects of hyperaridity in the interpretation of ?2Hwax values and pave the way for more quantitative paleohydrological reconstructions using ?2Hwax.

Background: Data are rare about the incidence of severe Fontan-associated liver disease (FALD) and its association with mortality. We sought to: (1) estimate the probability of developing severe FALD in patients who undergo the Fontan procedure (Fontan patients), compared with severe liver complications in patients with a ventricular septal defect; (2) assess the severe FALD-mortality association; and (3) identify risk factors for developing severe FALD. Methods and Results: Using the Quebec Congenital Heart Disease database, a total of 512 Fontan patients and 10232 patients with a ventricular septal defect were identified. Kaplan-Meier curves demonstrated significantly higher cumulative risk of severe FALD in Fontan patients (11.95% and 52.24% at 10 and 35years, respectively), than the risk of severe liver complica- tions in patients with a ventricular septal defect (0.50% and 2.75%, respectively). At 5years, the cumulative risk of death was 12.60% in patients with severe FALD versus 3.70% in Fontan patients without FALD (log-rank P=0.0171). Cox proportional hazard models identified significant associations between the development of severe FALD and congestive heart failure and supraventricular tachycardia, with hazard ratios (HRs) of 2.36 (95% CI, 1.38–4.02) and 2.45 (95% CI, 1.37–4.39), respectively. More recent Fontan completion was related to reduced risks of severe FALD, with an HR of 0.95 (95% CI, 0.93–0.97) for each more recent year. Conclusions: This large-scale population-based study documents that severe FALD in Fontan patients was associated with a >3-fold increase in mortality. The risk of FALD is time-dependent and can reach >50% by 35years after the Fontan operation. Conditions promoting poor Fontan hemodynamics were associated with severe FALD development.

Revealing the Mechanistic Details for the Selective Deoxygenation of Carboxylic Acids over Dynamic MoO3 Catalysts Abstract: The selective activation of renewable carboxylic acids could enable the formation of a variety of highly valuable renewable products, including surfactants, valuable dienes, and renewable hydrogen carriers. A kinetic study is performed to enhance understanding of the selective deoxygenation of carboxylic acid on promoted MoO3 at mild temperatures. Although several studies indicate that deoxygenation of oxygenated biomass-derived compounds on MoO3 occurs via a reverse Mars–van Krevelen mechanism, this study suggests that the deoxygenation of pentanoic acid (PA) on an oxygen vacancy can also be explained by a Langmuir–Hinshelwood mechanism. A detailed analysis of the experimental data indicates that the incorporation of Pt on MoO3 shifts the reaction order with respect to hydrogen from 1 to 0.5 at a low partial pressure of PA. We reveal that the rate-determining step (RDS) shifts upon the incorporation of Pt from H2 dissociation to H addition to adsorbed acid molecules. We further illustrate how the RDS can shift as a function of PA coverage. The inhibition effect of PA and its possible causes are discussed for both MoO3 and 0.05 wt % Pt/MoO3 catalysts. Here, we decouple promoter effects from the creation of highly active sites located at the Pt/MoO3 interface. The nature of the active site involved upon Pt incorporation is also studied by separating Pt from MoO3 at a controlled distance using carbon nanotubes as hydrogen bridges, confirming that the kinetically relevant role of Pt is to serve as a promoter of the MoO3. Resumen: La activación selectiva de ácidos carboxílicos renovables podría permitir la formación de una variedad de productos renovables altamente valiosos, incluyendo tensioactivos, dienos valiosos y portadores de hidrógeno renovable. Se lleva a cabo un estudio cinético para mejorar la comprensión de la desoxigenación selectiva de ácido carboxílico sobre MoO3 promovido a temperaturas moderadas. Aunque varios estudios indican que la desoxigenación de compuestos derivados de biomasa oxigenada sobre MoO3 ocurre a través de un mecanismo Mars-van Krevelen inverso, este estudio sugiere que la desoxigenación del ácido pentanoico (PA) en una vacante de oxígeno también puede explicarse mediante un mecanismo de Langmuir-Hinshelwood. Un análisis detallado de los datos experimentales indica que la incorporación de Pt en MoO3 cambia el orden de reacción con respecto al hidrógeno de 1 a 0.5 a baja presión parcial de PA. Revelamos que el paso determinante de la velocidad (PDV) cambia con la incorporación de Pt desde la disociación de H2 hasta la adición de H a las moléculas de ácido adsorbidas. Ilustramos además cómo el PDV puede cambiar en función de la cobertura de PA. Se discute el efecto de inhibición de PA y sus posibles causas tanto para MoO3 como para los catalizadores 0.05% en peso de Pt/MoO3. Aquí, se desvinculan los efectos del promotor de la creación de sitios altamente activos ubicados en la interfaz Pt/MoO3. También se estudia la naturaleza del sitio activo involucrado en la incorporación de Pt al separar Pt de MoO3 a una distancia controlada usando nanotubos de carbono como puentes de hidrógeno, confirmando que el papel cinéticamente relevante de Pt es servir como promotor del MoO3.

Abstract: Pharmaceutical Compounds (PCs), are synthetic substances intensively consumed by humans and continuously released into the urban wastewater. Often, the removal of PCs from wastewater is found to be insufficient, or even absent if population is not connected to the sewage system and wastewater is directly emitted into the environment. In the last decades, the PCs have aroused special attention because they have been found in surface water at concentrations that can cause negative effects on the aquatic biota and environment. Studies employing PCs monitoring and modelling techniques on surface waters have been mainly conducted in developed countries, while studies in developing countries are still scarce. Attempting to fill this gap, this work focused specifically in Colombia, where the occurrence of PCs has not been thoroughly studied and monitoring plans are barely done due to high costs. This work employed the Exposure to Pharmaceuticals in the Environment (ePiE), a spatially distributed model only implemented and used in Europe, to predict concentrations of PCs in rivers and lakes by considering their consumption, fate and transport. So, it aimed to create a new ePiE model network taking as a case of study the prediction of Naproxen and Paracetamol in the Upper Catchment of the Cauca River, Colombia, an area with a different socio-economic and climatic conditions with respect to the European continent. The methodological approach included: a) data collection on PCs’ consumption, physicochemical properties and removal in Wastewater Treatment Plants (WWTPs); b) consolidation of geodata of the study area such as hydrology, hydrography, urban agglomeration and WWTPs location; c) model building by using the open-source software environment R; d) analysis of the model sensitivity to input parameters; e) model validation. The results showed that all concentration predictions deviate within a range of only one order of magnitude from measurements which is a remarkable result when dealing with such high uncertainties. In terms of magnitude prediction, for the worst-case scenario, Naproxen and Paracetamol were found to exceed the environmental standards thresholds in 9% and 1% of the model river network nodes respectively. Overall, this research aimed to assess the applicability of the ePiE model in estimating the presence of PCs in a subregion of Colombia, demonstrating that it is feasible to expand ePiE to other areas and giving insights about the spatial distribution of pharmaceuticals substances so hotspots can be identified and several scenarios can be built in future works.

Resumen: El estudio resalta la oportunidad para fortalecer la biodiversidad, especialmente a través de la enseñanza de las ciencias ambientales y biociencias con un impacto esperado en el aumento en los niveles de alfabetización científica y su relación bidireccional con la educación y la cultura ambiental en la sociedad colombiana, con foco de estudio e impacto en la población estudiantil de ciencias naturales, biología y educación ambiental en los tres niveles educativos, así como la comunidad académica de las áreas de bioeconomía, ecosistemas naturales y territorios sostenibles. Se realizó una revisión y selección de la literatura científica bajo criterios de inclusión y exclusión como el año de publicación, los idiomas, los países de publicación y/o ejecución del estudio y las bases de datos utilizadas. Así, se tuvieron en cuenta autores que publicaran trabajos de investigación cualitativa y cuantitativa tanto aplicada como no aplicada tales como Boff, Caamaño, Calixto Flores, García & Rodríguez, González González y López & Sánchez. Se encontraron numerosas tendencias, tensiones y limitaciones en la integración de los conceptos de educación ambiental, biodiversidad y alfabetización científica, que se presentan como desafíos en el qué, el para qué y la estructura de la educación ambiental; las definiciones y la enseñanza en la alfabetización científica y la complejidad y relevancia en la comprensión de la biodiversidad. Se hallaron herramientas y conceptualizaciones con un impacto positivo para la conservación de los ecosistemas naturales estratégicos del país como la transversalización de la educación ambiental en el currículo; y el uso sostenible de los recursos por parte de las comunidades y la resignificación y apropiación del ambiente dentro de la cultura de las regiones como las necesidades ambientales contextuales de las poblaciones alfabetizadas. Se resalta el abordaje amplio y complejo de la biodiversidad dentro de un planteamiento nuevo hacia una pedagogía ambiental.

Abstract: With growing traffic demands and structural degradation accelerated by climate change, there is a critical need for continued advancement of concrete repair and strengthening technologies to enable extended bridge service life. Specifically, transitioning to cost-effective reinforced concrete (RC) retrofit strategies that further enhance durability under service loading conditions and minimize damage development under extreme hazard conditions, which are more probable to occur in longer-lasting concrete structures, are key elements of next-generation concrete bridges. This thesis explores the use of iron-based shape memory alloys (Fe-SMAs) as an active shear retrofit strategy for RC components. SMAs are a class of smart materials with the unique property of the shape memory effect, allowing them to fully recover plastic deformations when subsequently heated. By installing prestrained Fe-SMA strips and activating the shape memory effect, an active pressure can be introduced to help close cracks and apply a confining stress in the concrete. The primary objective is to evaluate the performance and practicality of Fe-SMA as an active shear strengthening technique in comparison to traditional methods such as external steel reinforcement and fibre-reinforced polymer (FRP) composites. The experimental phase of this study involved conducting push-off tests on RC specimens retrofitted with pre-deformed Fe-SMA strips. The goal was to assess the efficiency of active shear retrofitting. Additionally, finite element analysis (FEA) simulations were employed to model Fe SMA-retrofitted RC structures and investigate their behaviour under shear loading conditions. Key findings indicate that Fe-SMA retrofits, through transverse prestressing as part of the active retrofit strategy, contributed to notable improvements in the stiffness and strength of damaged RC components. Similar to passive retrofit methods, increased shear capacity was observed with higher levels of transverse reinforcement. Notably, combinations of substantial shear reinforcement ratios and elevated transverse prestressing provided the most significant gains in shear strength. Furthermore, the addition of prestressed Fe-SMA retrofits was found to effectively reduce shear crack widths and mitigate the progression of subsequent shear crack width growth. This study not only demonstrates the potential of Fe-SMA as a promising solution for active shear strengthening but also contributes to the development of future field-scale tests. The presence of Fe-SMA in damaged structures offers the prospect of multiple improvements, marking a significant advancement in the field of shear retrofitting.

Abstract: The use of computer vision, deep learning, and drones has revolutionized agriculture by enabling efficient crop monitoring and disease detection. Still, many challenges need to be overcome due to the vast diversity of plant species and their unique regional characteristics. Olive trees, which have been cultivated for thousands of years, present a particularly complex case for leaf-based disease diagnosis as disease symptoms can vary widely, both between different plant variations and even within individual leaves on the same plant. This complexity, coupled with the susceptibility of olive groves to various pathogens, including bacterial blight, olive knot, aculus olearius, and olive peacock spot, has hindered the development of effective disease detection algorithms. To address this challenge, we have devised a novel approach that combines deep learning techniques, leveraging convolutional neural networks, vision transformers, and cloud computing-based models. Aiming to detect and classify olive tree diseases the experimental results of our study have been highly promising, demonstrating the effectiveness of the combined transformer and cloud-based machine learning models, achieving an impressive accuracy of approximately 99.6% for multiclass classification cases including healthy, aculus olearius, and peacock spot infected leaves. These results highlight the potential of deep learning models in tackling the complex-ities of olive leaf disease detection and the need for further research in the field. Resumen: El uso de visión por computadora, aprendizaje profundo y drones ha revolucionado la agricultura al permitir un monitoreo eficiente de cultivos y la detección de enfermedades. Sin embargo, aún existen muchos desafíos debido a la vasta diversidad de especies vegetales y sus características regionales únicas. Los olivos, que han sido cultivados durante miles de años, presentan un caso especialmente complejo para el diagnóstico de enfermedades basado en las hojas, ya que los síntomas pueden variar ampliamente, tanto entre diferentes variaciones de plantas como incluso dentro de hojas individuales en la misma planta. Esta complejidad, junto con la susceptibilidad de los olivares a varios patógenos, como la bacteriosis, el nudo del olivo, el aculo olearius y la mancha de pavo real del olivo, ha dificultado el desarrollo de algoritmos efectivos de detección de enfermedades. Para abordar este desafío, hemos ideado un enfoque novedoso que combina técnicas de aprendizaje profundo, aprovechando redes neuronales convolucionales, transformadores de visión y modelos basados en computación en la nube. Con el objetivo de detectar y clasificar enfermedades de los olivos, los resultados experimentales de nuestro estudio han sido muy prometedores, demostrando la efectividad de los modelos combinados de transformadores y aprendizaje automático basado en la nube, logrando una impresionante precisión de aproximadamente el 99.6% para casos de clasificación multiclase, que incluyen hojas sanas, infectadas con aculo olearius y mancha de pavo real. Estos resultados destacan el potencial de los modelos de aprendizaje profundo para abordar las complejidades de la detección de enfermedades en las hojas de olivo y la necesidad de más investigaciones en este campo.

Abstract: Artificial Intelligence-based (AI) analysis of large, curated medical datasets is promising for providing early detection, faster diagnosis, and more effective treatment using low-power Electrocardiography (ECG) monitoring devices information. However, accessing sensitive medical data from diverse sources is highly restricted since improper use, unsafe storage, or data leakage could violate a person's privacy. This work uses a Federated Learning (FL) privacy-preserving methodology to train AI models over heterogeneous sets of high-definition ECG from 12-lead sensor arrays collected from six heterogeneous sources. We evaluated the capacity of the resulting models to achieve equivalent performance compared to state-of-the-art models trained in a Centralized Learning (CL) fashion. Moreover, we assessed the performance of our solution over Independent and Identical distributed (IID) and Non-IID federated data. Our methodology involves machine learning techniques based on Deep Neural Networks and Long-Short-Term Memory models. It has a robust data preprocessing pipeline with feature engineering, selection, and data balancing techniques. Our AI models demonstrated comparable performance to models trained using CL, IID, and Non-IID approaches. They showcased advantages in reduced complexity and faster training time, making them well-suited for cloud-edge architectures. Resumen: El análisis basado en Inteligencia Artificial (IA) de grandes conjuntos de datos médicos curados promete proporcionar detección temprana, diagnóstico más rápido y tratamiento más efectivo utilizando información de dispositivos de monitoreo de Electrocardiografía (ECG) de bajo consumo. Sin embargo, el acceso a datos médicos sensibles de diversas fuentes está altamente restringido, ya que el uso indebido, el almacenamiento inseguro o la fuga de datos podrían violar la privacidad de una persona. Este trabajo utiliza una metodología de Aprendizaje Federado (FL) preservadora de la privacidad para entrenar modelos de IA sobre conjuntos heterogéneos de ECG de alta definición provenientes de matrices de sensores de 12 derivaciones recopilados de seis fuentes heterogéneas. Evaluamos la capacidad de los modelos resultantes para lograr un rendimiento equivalente en comparación con modelos de vanguardia entrenados de manera centralizada (CL). Además, evaluamos el rendimiento de nuestra solución sobre datos federados Independientes e Idénticamente Distribuidos (IID) y No IID. Nuestra metodología involucra técnicas de aprendizaje automático basadas en Redes Neuronales Profundas y modelos de Memoria a Corto y Largo Plazo. Cuenta con un sólido pipeline de preprocesamiento de datos con técnicas de ingeniería de características, selección y equilibrio de datos. Nuestros modelos de IA demostraron un rendimiento comparable a los modelos entrenados mediante CL, IID y enfoques No IID. Exhibieron ventajas en términos de complejidad reducida y tiempo de entrenamiento más rápido, haciéndolos adecuados para arquitecturas de nube-edge.

Abstract: This dissertation explores how three applications of AI, namely classifiers, natural language processing and generative AI, have contributed and might contribute to the realisation of Richard Susskind’s predictions for the future of legal practice, as set in his 2023 book ‘Tomorrow’s Lawyers’. Besides reviewing Susskind’s predictions, it proposes a model to describe how lawyers work and uses it to identify existing uses of technology in the legal field. Also, it provides a general and non-technical approach to AI and its history and main features. With this context, it delves into the selected AI applications and explores how they work, their early legal applications, and their challenges. The research’s conclusion highlights the selected AI applications as powerful enablers for Susskind’s predictions. Moreover, how generative AI has the potential to disrupt the market of legal services and outperform lawyers in multiple tasks. A key insight refers to the latent market of underserved individuals, communities, and small and medium enterprises and how it offers an unbeatable opportunity for online-embedded generative AI-based legal services, which would improve access to legal services. Finally, this dissertation suggests an unavoidable shift in the general comprehension of the abbreviations ‘LL.M’ and ‘LLM’ within the legal field.

Resumen: El artículo presenta las conclusiones de la investigación doctoral en medios, comunicación y cultura, centrándose en la narrativa audiovisual de cuatro cortometrajes producidos por el colectivo de comunicación popular Alharaca por la paz y las acciones de cambio social producto de su movilización, con el fin de conocer cómo desde la comunicación comunitaria, local y participativa producida por niñas, niños y adolescentes de Caquetá, Colombia, se contribuye a la reconstrucción del tejido social en sociedades divididas por el conflicto armado. Abstract: The article presents the conclusions of the doctoral research in media, communication and culture, focusing on the audiovisual narrative of four short films produced by the popular communication collective Alharaca por la paz and the actions of social change resulting from its mobilization, in order to learn how community, local and participatory communication produced by girls, boys and adolescents from Caquetá, Colombia, contributes to the reconstruction of the social fabric in societies divided by the armed conflict.