Search

Your search keyword '"Tato, A."' showing total 8,368 results
Start Over Author "Tato, A."
8,368 results on '"Tato, A."'

1. On the locality bias and results in the Long Range Arena

Author

Miralles-González, Pablo, Huertas-Tato, Javier, Martín, Alejandro, and Camacho, David

Subjects
Computer Science - Computation and Language, Computer Science - Artificial Intelligence
Abstract

The Long Range Arena (LRA) benchmark was designed to evaluate the performance of Transformer improvements and alternatives in long-range dependency modeling tasks. The Transformer and its main variants performed poorly on this benchmark, and a new series of architectures such as State Space Models (SSMs) gained some traction, greatly outperforming Transformers in the LRA. Recent work has shown that with a denoising pre-training phase, Transformers can achieve competitive results in the LRA with these new architectures. In this work, we discuss and explain the superiority of architectures such as MEGA and SSMs in the Long Range Arena, as well as the recent improvement in the results of Transformers, pointing to the positional and local nature of the tasks. We show that while the LRA is a benchmark for long-range dependency modeling, in reality most of the performance comes from short-range dependencies. Using training techniques to mitigate data inefficiency, Transformers are able to reach state-of-the-art performance with proper positional encoding. In addition, with the same techniques, we were able to remove all restrictions from SSM convolutional kernels and learn fully parameterized convolutions without decreasing performance, suggesting that the design choices behind SSMs simply added inductive biases and learning efficiency for these particular tasks. Our insights indicate that LRA results should be interpreted with caution and call for a redesign of the benchmark.

Published
2025

2. Not all tokens are created equal: Perplexity Attention Weighted Networks for AI generated text detection

Author

Miralles-González, Pablo, Huertas-Tato, Javier, Martín, Alejandro, and Camacho, David

Subjects
Computer Science - Computation and Language, Computer Science - Artificial Intelligence
Abstract

The rapid advancement in large language models (LLMs) has significantly enhanced their ability to generate coherent and contextually relevant text, raising concerns about the misuse of AI-generated content and making it critical to detect it. However, the task remains challenging, particularly in unseen domains or with unfamiliar LLMs. Leveraging LLM next-token distribution outputs offers a theoretically appealing approach for detection, as they encapsulate insights from the models' extensive pre-training on diverse corpora. Despite its promise, zero-shot methods that attempt to operationalize these outputs have met with limited success. We hypothesize that one of the problems is that they use the mean to aggregate next-token distribution metrics across tokens, when some tokens are naturally easier or harder to predict and should be weighted differently. Based on this idea, we propose the Perplexity Attention Weighted Network (PAWN), which uses the last hidden states of the LLM and positions to weight the sum of a series of features based on metrics from the next-token distribution across the sequence length. Although not zero-shot, our method allows us to cache the last hidden states and next-token distribution metrics on disk, greatly reducing the training resource requirements. PAWN shows competitive and even better performance in-distribution than the strongest baselines (fine-tuned LMs) with a fraction of their trainable parameters. Our model also generalizes better to unseen domains and source models, with smaller variability in the decision boundary across distribution shifts. It is also more robust to adversarial attacks, and if the backbone has multilingual capabilities, it presents decent generalization to languages not seen during supervised training, with LLaMA3-1B reaching a mean macro-averaged F1 score of 81.46% in cross-validation with nine languages.

Published
2025

3. Isolating authorship from content with semantic embeddings and contrastive learning

Author

Huertas-Tato, Javier, Girón-Jiménez, Adrián, Martín, Alejandro, and Camacho, David

Subjects
Computer Science - Computation and Language, Computer Science - Machine Learning
Abstract

Authorship has entangled style and content inside. Authors frequently write about the same topics in the same style, so when different authors write about the exact same topic the easiest way out to distinguish them is by understanding the nuances of their style. Modern neural models for authorship can pick up these features using contrastive learning, however, some amount of content leakage is always present. Our aim is to reduce the inevitable impact and correlation between content and authorship. We present a technique to use contrastive learning (InfoNCE) with additional hard negatives synthetically created using a semantic similarity model. This disentanglement technique aims to distance the content embedding space from the style embedding space, leading to embeddings more informed by style. We demonstrate the performance with ablations on two different datasets and compare them on out-of-domain challenges. Improvements are clearly shown on challenging evaluations on prolific authors with up to a 10% increase in accuracy when the settings are particularly hard. Trials on challenges also demonstrate the preservation of zero-shot capabilities of this method as fine tuning.

Published
2024

4. A CLIP-based siamese approach for meme classification

Author

Huertas-Tato, Javier, Koutlis, Christos, Papadopoulos, Symeon, Camacho, David, and Kompatsiaris, Ioannis

Subjects
Computer Science - Multimedia
Abstract

Memes are an increasingly prevalent element of online discourse in social networks, especially among young audiences. They carry ideas and messages that range from humorous to hateful, and are widely consumed. Their potentially high impact requires adequate means of control to moderate their use in large scale. In this work, we propose SimCLIP a deep learning-based architecture for cross-modal understanding of memes, leveraging a pre-trained CLIP encoder to produce context-aware embeddings and a Siamese fusion technique to capture the interactions between text and image. We perform an extensive experimentation on seven meme classification tasks across six datasets. We establish a new state of the art in Memotion7k with a 7.25% relative F1-score improvement, and achieve super-human performance on Harm-P with 13.73% F1-Score improvement. Our approach demonstrates the potential for compact meme classification models, enabling accurate and efficient meme monitoring. We share our code at https://github.com/jahuerta92/meme-classification-simclip

Published
2024

5. DisTrack: a new Tool for Semi-automatic Misinformation Tracking in Online Social Networks

Author

Villar-Rodríguez, Guillermo, Huertas-García, Álvaro, Martín, Alejandro, Huertas-Tato, Javier, and Camacho, David

Subjects
Computer Science - Social and Information Networks, Computer Science - Artificial Intelligence
Abstract

Introduction: This article introduces DisTrack, a methodology and a tool developed for tracking and analyzing misinformation within Online Social Networks (OSNs). DisTrack is designed to combat the spread of misinformation through a combination of Natural Language Processing (NLP) Social Network Analysis (SNA) and graph visualization. The primary goal is to detect misinformation, track its propagation, identify its sources, and assess the influence of various actors within the network. Methods: DisTrack's architecture incorporates a variety of methodologies including keyword search, semantic similarity assessments, and graph generation techniques. These methods collectively facilitate the monitoring of misinformation, the categorization of content based on alignment with known false claims, and the visualization of dissemination cascades through detailed graphs. The tool is tailored to capture and analyze the dynamic nature of misinformation spread in digital environments. Results: The effectiveness of DisTrack is demonstrated through three case studies focused on different themes: discredit/hate speech, anti-vaccine misinformation, and false narratives about the Russia-Ukraine conflict. These studies show DisTrack's capabilities in distinguishing posts that propagate falsehoods from those that counteract them, and tracing the evolution of misinformation from its inception. Conclusions: The research confirms that DisTrack is a valuable tool in the field of misinformation analysis. It effectively distinguishes between different types of misinformation and traces their development over time. By providing a comprehensive approach to understanding and combating misinformation in digital spaces, DisTrack proves to be an essential asset for researchers and practitioners working to mitigate the impact of false information in online social environments.

Published
2024

6. Using Contrastive Learning to Map Stylistic Similarities in Narrative Writers

Author

Valero-Redondo, María, Huertas-Tato, Javier, Maceiras, Sergio D’Antonio, Martín, Alejandro, Camacho, David, Goos, Gerhard, Series Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Julian, Vicente, editor, Camacho, David, editor, Yin, Hujun, editor, Alberola, Juan M., editor, Nogueira, Vitor Beires, editor, Novais, Paulo, editor, and Tallón-Ballesteros, Antonio, editor

Published
2025
Full Text
View/download PDF

7. Multimodal Visio-Lingual Content Analysis to Detect Fake Content on Reddit

Author

Girón, Adrián, Huertas-Tato, Javier, Camacho, David, Goos, Gerhard, Series Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Julian, Vicente, editor, Camacho, David, editor, Yin, Hujun, editor, Alberola, Juan M., editor, Nogueira, Vitor Beires, editor, Novais, Paulo, editor, and Tallón-Ballesteros, Antonio, editor

Published
2025
Full Text
View/download PDF

10. Camouflage is all you need: Evaluating and Enhancing Language Model Robustness Against Camouflage Adversarial Attacks

Author

Huertas-García, Álvaro, Martín, Alejandro, Huertas-Tato, Javier, and Camacho, David

Subjects
Computer Science - Computation and Language
Abstract

Adversarial attacks represent a substantial challenge in Natural Language Processing (NLP). This study undertakes a systematic exploration of this challenge in two distinct phases: vulnerability evaluation and resilience enhancement of Transformer-based models under adversarial attacks. In the evaluation phase, we assess the susceptibility of three Transformer configurations, encoder-decoder, encoder-only, and decoder-only setups, to adversarial attacks of escalating complexity across datasets containing offensive language and misinformation. Encoder-only models manifest a 14% and 21% performance drop in offensive language detection and misinformation detection tasks, respectively. Decoder-only models register a 16% decrease in both tasks, while encoder-decoder models exhibit a maximum performance drop of 14% and 26% in the respective tasks. The resilience-enhancement phase employs adversarial training, integrating pre-camouflaged and dynamically altered data. This approach effectively reduces the performance drop in encoder-only models to an average of 5% in offensive language detection and 2% in misinformation detection tasks. Decoder-only models, occasionally exceeding original performance, limit the performance drop to 7% and 2% in the respective tasks. Although not surpassing the original performance, Encoder-decoder models can reduce the drop to an average of 6% and 2% respectively. Results suggest a trade-off between performance and robustness, with some models maintaining similar performance while gaining robustness. Our study and adversarial training techniques have been incorporated into an open-source tool for generating camouflaged datasets. However, methodology effectiveness depends on the specific camouflage technique and data encountered, emphasizing the need for continued exploration., Comment: 19 pages, 8 figures, 5 tables

Published
2024

15. Author Correction: Para-infectious brain injury in COVID-19 persists at follow-up despite attenuated cytokine and autoantibody responses

Author

Michael, Benedict D., Dunai, Cordelia, Needham, Edward J., Tharmaratnam, Kukatharmini, Williams, Robyn, Huang, Yun, Boardman, Sarah A., Clark, Jordan J., Sharma, Parul, Subramaniam, Krishanthi, Wood, Greta K., Collie, Ceryce, Digby, Richard, Ren, Alexander, Norton, Emma, Leibowitz, Maya, Ebrahimi, Soraya, Fower, Andrew, Fox, Hannah, Tato, Esteban, Ellul, Mark A., Sunderland, Geraint, Held, Marie, Hetherington, Claire, Egbe, Franklyn N., Palmos, Alish, Stirrups, Kathy, Grundmann, Alexander, Chiollaz, Anne-Cecile, Sanchez, Jean-Charles, Stewart, James P., Griffiths, Michael, Solomon, Tom, Breen, Gerome, Coles, Alasdair J., Kingston, Nathalie, Bradley, John R., Chinnery, Patrick F., Cavanagh, Jonathan, Irani, Sarosh R., Vincent, Angela, Baillie, J. Kenneth, Openshaw, Peter J., Semple, Malcolm G., Taams, Leonie S., and Menon, David K.

Published
2024
Full Text
View/download PDF

16. Follow-up for 3 years of a pediatric population diagnosed in 2018 with mother-to-child transmission of HIV in 8 Latin American countries in the PLANTAIDS cohort

Author

Vallejo, Beatriz Álvarez, Lobo, Alicia Hernanz, García, Itzíar Carrasco, Pérez, Tomás Bruno, Mino-Leon, Greta, Pazmiño, Judith Rosabel Soffe, Lorenzana, Julio Werner Juarez, Drummond, Tatiana, del Socorro Pavía Ruz, Noris Marlene, del Rocío Muñoz Hernández, María, Pérez, Dulce María Morales, Estripeaut, Dora, Luciani, Kathia, Martínez, Karen Sobeida Erazo, Villatoro, Luis Guillermo Castaneda, Madrigal, Oscar Porras, Ivankovich-Escoto, Gabriela, Tato, Luis Manuel Prieto, and Gómez, María Luisa Navarro

Published
2024
Full Text
View/download PDF

17. Estudio genético en adultos con glomeruloesclerosis focal y segmentaria

Author

Melissa Pilco-Terán, Amir Shabaka, Mónica Furlano, Ana Tato Ribera, Isabel Galán Carrillo, Eduardo Gutiérrez, Roser Torra, and Gema Fernández-Juárez

Subjects
Genetic, Focal Segmental Glomerulosclerosis, Indications, Diseases of the genitourinary system. Urology, RC870-923
Abstract

Resumen: La glomeruloesclerosis focal y segmentaria (GEFS) es un patrón histológico de lesión que deriva de diversos procesos patológicos que afectan a los podocitos, resultando en pérdida de selectividad del filtrado glomerular, proteinuria y desarrollando insuficiencia renal que progresa a enfermedad renal crónica terminal en un importante número de pacientes. La clasificación propuesta por las guías KDIGO, en 2021, divide la GEFS en cuatro categorías: primaria, secundaria, genética y de causa no determinada, facilitando así su diagnóstico y manejo. Las causas hereditarias de la GEFS presentan una variabilidad clínica significativa, complicando su identificación. El estudio genético es crucial para identificar la GEFS de causa genética. La prevalencia de la GEFS genética es significativa en niños y considerable en adultos, destacando la importancia del diagnóstico temprano para evitar tratamientos innecesarios y facilitar el consejo genético. Las técnicas de secuenciación masiva han revolucionado el diagnóstico genético, permitiendo la identificación de más de 60 genes responsables del daño podocitario. Este documento propone recomendaciones clínicas y patológicas para la realización de estudios genéticos en adultos con GEFS, subrayando la necesidad de una correcta clasificación para la planificación terapéutica adecuada y la mejora de los resultados en ensayos clínicos. Abstract: Focal segmental glomerulosclerosis (FSGS) is a histological pattern of injury that derives from various pathological processes that affect podocytes, resulting in loss of selectivity of the glomerular filtration membrane, proteinuria and the development of renal failure that progresses to end-stage kidney disease in a significant number of patients. The classification proposed by the 2021 KDIGO guidelines divides FSGS into four categories: primary, secondary, genetic, and FSGS of undetermined cause, thus facilitating its diagnosis and management. Genetic causes of FSGS present significant clinical variability, complicating their identification. Genetic testing is crucial to identify FSGS of genetic cause. The prevalence of genetic FSGS is significant in children and considerable in adults, highlighting the importance of early diagnosis to avoid unnecessary treatments and facilitate genetic counselling. Massive sequencing techniques have revolutionized genetic diagnosis, allowing the identification of more than 60 genes responsible for podocyte damage. This document proposes clinical recommendations for carrying out genetic studies in adults with FSGS, highlighting the need for a correct classification for adequate therapeutic planning and improvement of results in clinical trials.

Published
2025
Full Text
View/download PDF

18. Leveraging culinary nutrition interventions and contextual factors to combat child growth failure in mining communities for maximum impact

Author

Herbert Tato Nyirenda, David Mulenga, and Hilda Nyambe-Silavwe

Subjects
Stunting, Under-five, Culinary nutrition, Combined effect, Growth failure, Nutrition. Foods and food supply, TX341-641
Abstract

Summary: Background and aims: A growing body of evidence supports the effectiveness of culinary nutrition interventions; however, there is limited research on the analytical techniques, components, and the combined impact of these interventions alongside contextual factors in addressing child growth failure. This paper examines the combined effect of culinary nutrition interventions and contextual factors in reducing child growth failure within mining communities. Methods and materials: This study employed a cross-sectional design, targeting children under the age of five and their caregivers. A total of 706 participants were included, selected using a systematic random sampling technique. Data were collected through structured questionnaires, direct child measurements, and structured interviews. Anthropometric data were analysed using the WHO growth standards. Statistical analysis involved descriptive statistics and logistic regression, with an emphasis on examining interaction effects. Results: The results revealed that 34.7% of the children experienced growth failure while 55% of the caregivers participated in culinary nutrition programs. Children whose caregivers participated in culinary programs had significantly lower odds of experiencing growth failure (OR = 0.69∗∗, P = 0.019, 95% CI). Factorial analysis demonstrated that the impact of culinary interventions on reducing child growth failure was more pronounced in households with adequate food (OR = 0.46∗, P = 0.085, 90% CI), households with vegetable gardens (OR = 0.50∗∗, P = 0.005, 95% CI), children with a good appetite (OR = 0.44∗∗, P = 0.032, 95% CI), households practicing food preservation (OR = 0.52∗, P = 0.054, 90% CI), and those producing their own food (OR = 0.46∗∗, P = 0.016, 95% CI). Conclusion: The findings highlight the synergistic impact of culinary nutrition interventions alongside key factors in addressing child growth failure, particularly in mining communities. It emphasizes the importance of integrating both nutritional programs and the social environment to effectively combat child malnutrition and stunting in these vulnerable populations.

Published
2025
Full Text
View/download PDF

19. Bio-Inspired Algorithms in Healthcare

Author

Firdaws Rizgar Tato and Ibrahim Mahmood Ibrahim

Subjects
orthogonal the local preserving projection, colony, medical the information classification, training data, testing data, Information technology, T58.5-58.64, Computer software, QA76.75-76.765
Abstract

Exploring hidden patterns in medical data sets is made possible by the huge potential of medical data mining. A clinical diagnosis can be made with the help of these patterns. Research on bio-inspired algorithms is a recent development. Its primary benefit is its ability to weave together social behavior, emergence, and connectionism subfields. In a nutshell, it involves modeling live phenomena using computers while studying life to make better computer applications. This chapter describes the application of five bio-inspired algorithms, including metaheuristics, to classify seven distinct real health-related information sets. While the other two of these methods rely on random population creation to create classification rules, the other two rely on the computation of similarity between the data used for training and testing. The outcomes demonstrated that bio-inspired supervised medical data classification methods were incredibly effective.

Published
2024
Full Text
View/download PDF

20. Understanding writing style in social media with a supervised contrastively pre-trained transformer

Author

Huertas-Tato, Javier, Martin, Alejandro, and Camacho, David

Subjects
Computer Science - Computation and Language, Computer Science - Social and Information Networks
Abstract

Online Social Networks serve as fertile ground for harmful behavior, ranging from hate speech to the dissemination of disinformation. Malicious actors now have unprecedented freedom to misbehave, leading to severe societal unrest and dire consequences, as exemplified by events such as the Capitol assault during the US presidential election and the Antivaxx movement during the COVID-19 pandemic. Understanding online language has become more pressing than ever. While existing works predominantly focus on content analysis, we aim to shift the focus towards understanding harmful behaviors by relating content to their respective authors. Numerous novel approaches attempt to learn the stylistic features of authors in texts, but many of these approaches are constrained by small datasets or sub-optimal training losses. To overcome these limitations, we introduce the Style Transformer for Authorship Representations (STAR), trained on a large corpus derived from public sources of 4.5 x 10^6 authored texts involving 70k heterogeneous authors. Our model leverages Supervised Contrastive Loss to teach the model to minimize the distance between texts authored by the same individual. This author pretext pre-training task yields competitive performance at zero-shot with PAN challenges on attribution and clustering. Additionally, we attain promising results on PAN verification challenges using a single dense layer, with our model serving as an embedding encoder. Finally, we present results from our test partition on Reddit. Using a support base of 8 documents of 512 tokens, we can discern authors from sets of up to 1616 authors with at least 80\% accuracy. We share our pre-trained model at huggingface (https://huggingface.co/AIDA-UPM/star) and our code is available at (https://github.com/jahuerta92/star)

Published
2023

21. Whole-genome sequencing rule-out of suspected hospital-onset Rhizopus outbreaks

Author

Chu, Victoria T, Nafees, Saba, Waltari, Eric, McNeil, Nicole, Caughell, Carolyn, Sanchez-Guerrero, Estella, Wang, Lusha, Stanley, Kim, Cunningham, Gail, Wong, Joan, Phelps, Maíra, Tato, Cristina M, Miller, Steve, DeRisi, Joseph L, Yokoe, Deborah S, Ramirez-Avila, Lynn, and Langelier, Charles R

Subjects
Biomedical and Clinical Sciences, Clinical Sciences, Infectious Diseases, Biotechnology, Clinical Research, Genetics, Human Genome, 4.2 Evaluation of markers and technologies, Infection, Humans, Rhizopus, Phylogeny, Genome, Bacterial, Hospitals, Disease Outbreaks, Medical and Health Sciences, Epidemiology, Biomedical and clinical sciences, Health sciences
Abstract

Two independent temporal-spatial clusters of hospital-onset Rhizopus infections were evaluated using whole-genome sequencing (WGS). Phylogenetic analysis confirmed that isolates within each cluster were unrelated despite epidemiological suspicion of outbreaks. The ITS1 region alone was insufficient for accurate analysis. WGS has utility for rapid rule-out of suspected nosocomial Rhizopus outbreaks.

Published
2023

22. Open-VERSO: a vision of 5G experimentation infrastructures, hurdles and challenges

Author

Martin, Angel, Losada, Pablo, Fernández, Carolina, Zorrilla, Mikel, Fernandez, Zaloa, Gabilondo, Alvaro, Uriol, Juncal, Mogollon, Felipe, Serón, Mikel, Dalgitsis, Michalis, Viola, Roberto, Roca, Luis, Giraldo, Carlos, Gonzalez, Pablo, Tato, Anxo, Escudero, Joaquín, Vazquez, Alvaro, Camps, Daniel, Cárdenas, Andrés, Herranz, Carlos, Aleixendri, Joan Josep, Iglesias, Rebeca, Cernigliaro, Gianluca, Montagud, Mario, Tomàs, Pau, and Giménez, Sergio

Subjects
Computer Science - Networking and Internet Architecture
Abstract

5G led to a digital revolution for networks by leveraging virtualisation techniques to manage software-based network functions through provided standard interfaces, which have matured recently for cloud infrastructure that is widely employed across domains and sectors. This undiscovered potential to adequately respond to concurrent and specialised traffic demands is promising for a wide spectrum of industries. Moreover, it exposes the networking ecosystem to prospects beyond the traditional value chain. However, the configuration, deployment and operation of a 5G network are challenging. Thus, different scientific and research entities have built their own open, evolvable and updateable testbed infrastructure that can be used for experimentation purposes. Such testbeds enable different stakeholders to integrate new systems or features exploiting new technologies, assess the performance of innovative services, and customise operation policies to find optimal setups from a cost-effective perspective. Furthermore, federations of infrastructure allow for wider and more complex experiments to be performed in distributed domains. However, numerous technical and procedural obstacles exist during the building of 5G network testbeds. In addition, some technical barriers persist despite the testing of alternatives and ongoing efforts within open-source systems and commercial equipment portfolios. All these limitations and challenges are relevant for experimenters and stakeholders as they attempt to determine the scope of 5G set expectations., Comment: 41 pages, 8 tables, White Paper

Published
2023

23. Spain on Fire: A novel wildfire risk assessment model based on image satellite processing and atmospheric information

Author

Liz-López, Helena, Huertas-Tato, Javier, Pérez-Aracil, Jorge, Casanova-Mateo, Carlos, Sanz-Justo, Julia, and Camacho, David

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Artificial Intelligence, Electrical Engineering and Systems Science - Image and Video Processing
Abstract

Each year, wildfires destroy larger areas of Spain, threatening numerous ecosystems. Humans cause 90% of them (negligence or provoked) and the behaviour of individuals is unpredictable. However, atmospheric and environmental variables affect the spread of wildfires, and they can be analysed by using deep learning. In order to mitigate the damage of these events we proposed the novel Wildfire Assessment Model (WAM). Our aim is to anticipate the economic and ecological impact of a wildfire, assisting managers resource allocation and decision making for dangerous regions in Spain, Castilla y Le\'on and Andaluc\'ia. The WAM uses a residual-style convolutional network architecture to perform regression over atmospheric variables and the greenness index, computing necessary resources, the control and extinction time, and the expected burnt surface area. It is first pre-trained with self-supervision over 100,000 examples of unlabelled data with a masked patch prediction objective and fine-tuned using 311 samples of wildfires. The pretraining allows the model to understand situations, outclassing baselines with a 1,4%, 3,7% and 9% improvement estimating human, heavy and aerial resources; 21% and 10,2% in expected extinction and control time; and 18,8% in expected burnt area. Using the WAM we provide an example assessment map of Castilla y Le\'on, visualizing the expected resources over an entire region.

Published
2023

24. Experience of the national cohort of pregnant women with HIV and their children in Spain: temporal trends in vertical transmission of HIV and associated infections

Author

Marta Illán Ramos, Arantxa Berzosa Sánchez, Itziar Carrasco García, Asunción Diaz Franco, Inmaculada Jarrín Vera, Luis Prieto Tato, Rosa Polo Rodríguez, Mª Luisa Navarro Gómez, and José Tomás Ramos Amador

Subjects
VIH, Transmisión vertical, Infección perinatal, Tratamiento antirretroviral, Pediatrics, RJ1-570
Abstract

Introduction: The vertical transmission rate (VTR) of HIV has decreased to less than 2% in high-income countries, in spite of which perinatal infections continue to occur. We present data from the national cohort of pregnant women living with HIV and their children in Spain. The objectives were to describe the characteristics of this population, evaluate the VTR of HIV, the safety of antiretroviral therapy (ART) and the prevalence of coinfection. Patients and methods: Multicentre prospective, observational and descriptive study with participation of 62 hospitals. The sample included pegnant women living with HIV whose children were born between January 2020 and December 2022. We collected prospective data on the characteristics of mothers and children using an online questionnaire (REDCap web application). Results: The study included 414 mother-child dyads. Most mothers were immigrants (227/349; 65.1%). The main route of HIV infection was heterosexual transmission (160/402; 39.8%), followed by vertical transmission (44/402; 10.9%). The diagnosis was made before conception in 313/389 women (80.4%), 394/402 (98%) received ART during pregnancy and 356/402 (89.3%) had an undetectable viral load at the time of delivery. The delivery was vaginal in 230/388 children (59.3%). The proportion of preterm birth was 11.1%. The most frequent neonatal prophylaxis approach was monotherapy with zidovudine (358/414; 86.5%). There were 3 cases of vertical transmission of HIV (95% CI, 0%–1.54%). Only one newborn was breastfed. Conclusions: At present, most women living with HIV in Spain receive the diagnosis before conception, are of foreign ancestry and achieve good control of the infection. Although the VTR is very low in Spain, there are still infections that could be prevented with early diagnosis and treatment. Resumen: Introducción: La tasa de transmisión materno-infantil (TMI) del VIH ha disminuido a menos del 2% en países desarrollados, a pesar de lo cual siguen produciéndose infecciones. Se presentan los datos de la Cohorte Nacional Española de mujeres embarazadas que viven con VIH y sus hijos. Los objetivos fueron describir sus características, evaluar la TMI de VIH, la seguridad del tratamiento antirretroviral (TAR) y la prevalencia de coinfecciones. Pacientes y métodos: Estudio descriptivo, observacional, prospectivo y multicéntrico con participación de 62 hospitales. Se incluyeron las embarazadas cuyos hijos nacieron entre enero 2020 y diciembre 2022. Se recogieron prospectivamente características de las madres y niños en REDCap. Resultados: Hubo 414 pares madre-niño. Las madres fueron migrantes 227/349 (65,1%). La principal vía de adquisición del VIH fue la heterosexual (160/402; 39,8%), seguido de la TMI (44/402; 10,9%). En 313/389 (80,4%) el diagnóstico fue previo al embarazo, 394/402 (98%) recibieron TAR durante la gestación y 356/402 (89,3%) presentaba carga viral indetectable al parto. Hubo 230/388 (59,3%) neonatos nacidos por parto vaginal. La tasa de prematuridad fue 11,1%. La profilaxis neonatal más empleada (358/414; 86,5%) fue monoterapia con zidovudina. Hubo 3 casos de TMI del VIH (0,72% (IC 95%; 0%–1,54%), todos ellos con infección intraútero. Todos menos 1 recibieron lactancia artificial. Conclusiones: Actualmente las mujeres que viven con VIH en nuestro medio son migrantes, diagnosticadas mayoritariamente antes de la gestación, y logran adecuada situación inmunovirológica. Aunque la tasa de TMI es muy baja en nuestro país continúan ocurriendo infecciones prevenibles con diagnóstico y tratamiento más precoces.

Published
2024
Full Text
View/download PDF

25. Multimodal Endoscopic Management of Fistula after Sleeve Gastrectomy Involving a Novel Esophageal Stent

Author

Claudia Fortunato, Carlos Noronha Ferreira, Miguel Moura, Amélia Almeida, Rui Tato Marinho, and Luís Correia

Subjects
vertical gastrectomy, fistulas, endoscopic management, stents, gastrectomia vertical, fístulas, manejo endoscópico, prótese, Diseases of the digestive system. Gastroenterology, RC799-869
Abstract

Introduction: Laparoscopic sleeve gastrectomy (LSG) is a bariatric surgical procedure associated with a non-negligible risk of postoperative adverse events, especially fistulas, the majority of which occur at the angle of His. This adverse event requires a multidisciplinary approach involving intensive care, surgery, interventional endoscopy, and radiology. Despite the absence of an algorithmic endoscopic approach, a primarily endoscopic management of fistulas after LSG is now standard of care in most institutions. Case Report: A 66-year-old female with grade III obesity, obstructive sleep apnea, type 2 diabetes, and hypertension underwent LSG. She developed abdominal pain, hypovolemic shock, and severe anemia (Hb 6.5 g/dL). A computed tomography (CT) scan revealed hemoperitoneum without active bleeding, managed with transfusion of packed blood cells. A week later, a new CT scan performed for leukocytosis and abdominal pain revealed pneumoperitoneum. An esophagogastroduodenoscopy revealed a 20-mm fistula orifice at the angle of His. A novel esophageal covered metallic stent was placed for a period of 5 weeks. The fistula orifice decreased to 4 mm and communicated through a fistulous tract with a residual subphrenic abscess measuring 62 × 20 mm. Pus was collected from the abscess and drained internally with a 10-Fr double pigtail plastic stent through the fistula orifice. Following an initial period of improvement, clinical deterioration required percutaneous subphrenic abscess drainage. Two weeks later, the double pigtail plastic stent was removed, the fistula orifice was ablated with argon plasma 40W/1L and closed with an over-the-scope clip of 10 mm. Patient improved and was discharged 4 months after the LSG. Conclusion: The Luso-Cor esophageal stent is a specifically designed covered metallic stent with a 5-mm uncovered ring near the proximal edge, which reduces the risk of migration. Two articulating zones in the middle portion allow better adaptation to altered anatomy after LSG and a distal flare reduces retrograde reflux of fluid. This stent overcomes strictures in the gastric tube, concomitantly present in nearly 50% of patients with fistulas after LSG. The novel Luso-Cor esophageal stent provided a bridge to clinical stability with a significant reduction in the size of the fistula orifice which was closed with complementary therapeutic endoscopic procedures.

Published
2024
Full Text
View/download PDF

26. Colaboração: 16 variações sobre um tema

Author

Maria Alice Poppe and Tato Taborda

Subjects
Contemporary Dance, Contemporary Music, Gesture, Silence, Concept, Drama, PN1600-3307, Dramatic representation. The theater, PN2000-3307
Abstract

The text investigates the interactions between dance and music in the piece Colaboração: 16 variações sobre um tema (Collaboration: 16 variations on a theme) by the dancer Maria Alice Poppe and the musician Tato Taborda. Its structure, a set of variations on a theme that reappears at the end, echoes Bach’s Goldberg Variations. Each of the variations refers to seminal themes, works and issues in contemporary dance and contemporary music, such as silence, noise, listening and body awareness, deepening the reflection on the interaction between sound and gesture, in the works and concepts of others artists, but also in works by the two artists, created over 10 years of artistic collaboration.

Published
2024

27. Colaboração ou 16 Variações sobre um Tema

Author

Maria Alice Poppe and Tato Taborda

Subjects
Dança Contemporânea, Música Contemporânea, Gesto, Silêncio, Conceito, Drama, PN1600-3307, Dramatic representation. The theater, PN2000-3307
Abstract

Colaboração ou 16 Variações sobre um Tema: conformações paradoxais entre dança, música e conceito – O texto investiga a relação entre dança e música na peça Colaboração: 16 variações sobre um tema, criada e interpretada pela bailarina Maria Alice Poppe e pelo músico Tato Taborda. Sua estrutura, um conjunto de variações sobre um tema que reaparece ao final, ecoa as Variações Goldberg de Bach. As variações de Colaboração remetem a temas e questões seminais da dança e da música contemporânea, como o silêncio, o ruído, a escuta e consciência corporal, aprofundando uma reflexão acerca da interação entre som e gesto nas obras e conceitos de outros criadores, mas, também, nas obras dos dois artistas criadas ao longo de 10 anos de colaboração artística.

Published
2024

28. Kohay, senpai e a escuta como método

Author

Tato Taborda

Subjects
Escuta, Sonologia, Acusmática, Composição musical, Fine Arts, Arts in general, NX1-820, Special aspects of education, LC8-6691
Abstract

Os termos senpai e kohai referem-se a um modo de relação entre estudantes onde o mais experiente, o senpai, serve de referência para o menos graduado. Esse texto, escrito pela perspectiva de um kohay, busca apresentar de forma breve alguns aspectos da atuação do compositor Rodolfo Caesar, seja no âmbito da criação musical ou em seu modo de navegação pelas vias sinuosas da vida acadêmica, acentuando a singularidade de sua produção nos dois campos. Partindo de sua formação em ambientes fortemente marcados pela experimentação, no Brasil e na França, o pensamento de Caesar sobre a escuta, posto em ato em obras e textos ao longo de mais de 4 décadas, impõe-se por sua originalidade e elegância.

Published
2024
Full Text
View/download PDF

29. Countering Malicious Content Moderation Evasion in Online Social Networks: Simulation and Detection of Word Camouflage

Author

Huertas-García, Álvaro, Martín, Alejandro, Tato, Javier Huertas, and Camacho, David

Subjects
Computer Science - Computation and Language, Computer Science - Artificial Intelligence, Computer Science - Social and Information Networks
Abstract

Content moderation is the process of screening and monitoring user-generated content online. It plays a crucial role in stopping content resulting from unacceptable behaviors such as hate speech, harassment, violence against specific groups, terrorism, racism, xenophobia, homophobia, or misogyny, to mention some few, in Online Social Platforms. These platforms make use of a plethora of tools to detect and manage malicious information; however, malicious actors also improve their skills, developing strategies to surpass these barriers and continuing to spread misleading information. Twisting and camouflaging keywords are among the most used techniques to evade platform content moderation systems. In response to this recent ongoing issue, this paper presents an innovative approach to address this linguistic trend in social networks through the simulation of different content evasion techniques and a multilingual Transformer model for content evasion detection. In this way, we share with the rest of the scientific community a multilingual public tool, named "pyleetspeak" to generate/simulate in a customizable way the phenomenon of content evasion through automatic word camouflage and a multilingual Named-Entity Recognition (NER) Transformer-based model tuned for its recognition and detection. The multilingual NER model is evaluated in different textual scenarios, detecting different types and mixtures of camouflage techniques, achieving an overall weighted F1 score of 0.8795. This article contributes significantly to countering malicious information by developing multilingual tools to simulate and detect new methods of evasion of content on social networks, making the fight against information disorders more effective., Comment: 21 pages, 4 figures, 6 tables, Fast Track from IDEAL 21 Conference

Published
2022

31. Ocean Species Discoveries 1–12 — A primer for accelerating marine invertebrate taxonomy

Author

Senckenberg Ocean Species Alliance (SOSA), Angelika Brandt, Chong Chen, Laura Engel, Patricia Esquete, Tammy Horton, Anna Jażdżewska, Nele Johannsen, Stefanie Kaiser, Terue Kihara, Henry Knauber, Katharina Kniesz, Jannes Landschoff, Anne-Nina Lörz, Fabrizio Machado, Carlos Martínez-Muñoz, Torben Riehl, Amanda Serpell-Stevens, Julia Sigwart, Anne Helene Tandberg, Ramiro Tato, Miwako Tsuda, Katarzyna Vončina, Hiromi Watanabe, Christian Wenz, and Jason Williams

Subjects
new species, shelf-life, Mollusca, Arthropoda, Ech, Biology (General), QH301-705.5
Abstract

Discoveries of new species often depend on one or a few specimens, leading to delays as researchers wait for additional context, sometimes for decades. There is currently little professional incentive for a single expert to publish a stand-alone species description. Additionally, while many journals accept taxonomic descriptions, even specialist journals expect insights beyond the descriptive work itself. The combination of these factors exacerbates the issue that only a small fraction of marine species are known and new discoveries are described at a slow pace, while they face increasing threats from accelerating global change. To tackle this challenge, this first compilation of Ocean Species Discoveries (OSD) presents a new collaborative framework to accelerate the description and naming of marine invertebrate taxa that can be extended across all phyla. Through a mode of publication that can be speedy, taxonomy-focused and generate higher citation rates, OSD aims to create an attractive home for single species descriptions. This Senckenberg Ocean Species Alliance (SOSA) approach emphasises thorough, but compact species descriptions and diagnoses, with supporting illustrations and with molecular data when available. Even basic species descriptions carry key data for distributions and ecological interactions (e.g., host-parasite relationships) besides universally valid species names; these are essential for downstream uses, such as conservation assessments and communicating biodiversity to the broader public.This paper presents thirteen marine invertebrate taxa, comprising one new genus, eleven new species and one re-description and reinstatement, covering wide taxonomic, geographic, bathymetric and ecological ranges. The taxa addressed herein span three phyla (Mollusca, Arthropoda, Echinodermata), five classes, eight orders and twelve families. Apart from the new genus, an updated generic diagnosis is provided for four other genera. The newly-described species of the phylum Mollusca are Placiphorella methanophila Vončina, sp. nov. (Polyplacophora, Mopaliidae), Lepetodrilus marianae Chen, Watanabe & Tsuda, sp. nov. (Gastropoda, Lepetodrilidae), Shinkailepas gigas Chen, Watanabe & Tsuda, sp. nov. (Gastropoda, Phenacolepadidae) and Lyonsiella illaesa Machado & Sigwart, sp. nov. (Bivalvia, Lyonsiellidae). The new taxa of the phylum Arthropoda are all members of the subphylum Crustacea: Lepechinella naces Lörz & Engel, sp. nov. (Amphipoda, Lepechinellidae), Cuniculomaera grata Tandberg & Jażdżewska, gen. et sp. nov. (Amphipoda, Maeridae), Pseudionella pumulaensis Williams & Landschoff, sp. nov. (Isopoda, Bopyridae), Mastigoniscus minimus Wenz, Knauber & Riehl, sp. nov. (Isopoda, Haploniscidae), Macrostylis papandreas Jonannsen, Riehl & Brandt, sp. nov. (Isopoda, Macrostylidae), Austroniscus indobathyasellus Kaiser, Kniesz & Kihara, sp. nov. (Isopoda, Nannoniscidae) and Apseudopsis daria Esquete & Tato, sp. nov. (Tanaidacea, Apseudidae). In the phylum Echinodermata, the reinstated species is Psychropotes buglossa E. Perrier, 1886 (Holothuroidea, Psychropotidae).The study areas span the North and Central Atlantic Ocean, the Indian Ocean and the North, East and West Pacific Ocean and depths from 5.2 m to 7081 m. Specimens of eleven free-living and one parasite species were collected from habitats ranging from an estuary to deep-sea trenches. The species were illustrated with photographs, line drawings, micro-computed tomography, confocal laser scanning microscopy and scanning electron microscopy images. Molecular data are included for nine species and four species include a molecular diagnosis in addition to their morphological diagnosis.The five new geographic and bathymetric distribution records comprise Lepechinella naces Lörz & Engel, sp. nov., Cuniculomaera grata Tandberg & Jażdżewska, sp. nov., Pseudionella pumulaensis Williams & Landschoff, sp. nov., Austroniscus indobathyasellus Kaiser, Kniesz & Kihara, sp. nov. and Psychropotes buglossa E. Perrier, 1886, with the novelty spanning from the species to the family level. The new parasite record is Pseudionella pumulaensis Williams & Landschoff, sp. nov., found in association with the hermit crab Pagurus fraserorum Landschoff & Komai, 2018.

Published
2024
Full Text
View/download PDF

32. PART: Pre-trained Authorship Representation Transformer

Author

Huertas-Tato, Javier, Huertas-Garcia, Alvaro, Martin, Alejandro, and Camacho, David

Subjects
Computer Science - Computation and Language
Abstract

Authors writing documents imprint identifying information within their texts: vocabulary, registry, punctuation, misspellings, or even emoji usage. Finding these details is very relevant to profile authors, relating back to their gender, occupation, age, and so on. But most importantly, repeating writing patterns can help attributing authorship to a text. Previous works use hand-crafted features or classification tasks to train their authorship models, leading to poor performance on out-of-domain authors. A better approach to this task is to learn stylometric representations, but this by itself is an open research challenge. In this paper, we propose PART: a contrastively trained model fit to learn \textbf{authorship embeddings} instead of semantics. By comparing pairs of documents written by the same author, we are able to determine the proprietary of a text by evaluating the cosine similarity of the evaluated documents, a zero-shot generalization to authorship identification. To this end, a pre-trained Transformer with an LSTM head is trained with the contrastive training method. We train our model on a diverse set of authors, from literature, anonymous blog posters and corporate emails; a heterogeneous set with distinct and identifiable writing styles. The model is evaluated on these datasets, achieving zero-shot 72.39\% and 86.73\% accuracy and top-5 accuracy respectively on the joint evaluation dataset when determining authorship from a set of 250 different authors. We qualitatively assess the representations with different data visualizations on the available datasets, profiling features such as book types, gender, age, or occupation of the author.

Published
2022

33. MultiSero: An Open-Source Multiplex-ELISA Platform for Measuring Antibody Responses to Infection.

Author

Byrum, Janie, Waltari, Eric, Janson, Owen, Guo, Syuan-Ming, Folkesson, Jenny, Chhun, Bryant, Vinden, Joanna, Ivanov, Ivan, Forst, Marcus, Li, Hongquan, Larson, Adam, Blackmon, Lena, Liu, Ziwen, Wu, Wesley, Ahyong, Vida, Tato, Cristina, McCutcheon, Krista, Hoh, Rebecca, Kelly, J, Martin, Jeffrey, Peluso, Michael, Henrich, Timothy, Deeks, Steven, Prakash, Manu, Greenhouse, Bryan, Mehta, Shalin, and Pak, John

Subjects
ELISA, SARS-CoV-2, multiplex, open-source, serology, serosurveillance
Abstract

A multiplexed enzyme-linked immunosorbent assay (ELISA) that simultaneously measures antibody binding to multiple antigens can extend the impact of serosurveillance studies, particularly if the assay approaches the simplicity, robustness, and accuracy of a conventional single-antigen ELISA. Here, we report on the development of multiSero, an open-source multiplex ELISA platform for measuring antibody responses to viral infection. Our assay consists of three parts: (1) an ELISA against an array of proteins in a 96-well format; (2) automated imaging of each well of the ELISA array using an open-source plate reader; and (3) automated measurement of optical densities for each protein within the array using an open-source analysis pipeline. We validated the platform by comparing antibody binding to Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) antigens in 217 human sera samples, showing high sensitivity (0.978), specificity (0.977), positive predictive value (0.978), and negative predictive value (0.977) for classifying seropositivity, a high correlation of multiSero determined antibody titers with commercially available SARS-CoV-2 antibody tests, and antigen-specific changes in antibody titer dynamics upon vaccination. The open-source format and accessibility of our multiSero platform can contribute to the adoption of multiplexed ELISA arrays for serosurveillance studies, for SARS-CoV-2 and other pathogens of significance.

Published
2023

40. Reputational Leadership, Family Business and Sustainability in the Creation of Economic and Social Value. The Case of the Company 'Agroparque Esperanza'

Author

Tato Jiménez, Juan Luis, Cabrera Gala, Ramsés, Carrillo Durán, Maria Victoria, Leitão, João, Series Editor, Devezas, Tessaleno, Series Editor, De Groot, (Bert) EA, Editorial Board Member, Ferreira, Joao, Editorial Board Member, Grubler, Arnulf, Editorial Board Member, LePoire, David, Editorial Board Member, De Oliveira Matias, Joao Carlos, Editorial Board Member, Rego, Conceição, editor, Lucas, Maria Raquel, editor, Sánchez-Hernández, María Isabel, editor, Cagica Carvalho, Luísa, editor, and Backx Noronha, Adriana, editor

Published
2024
Full Text
View/download PDF

41. Adapting Graphic Expression Teaching During the Pandemic: A Turning Point

Author

Conde-Fernández, Adriana, Tato-Sánchez-Del-Valle, Patricia Eva, Vila-Lameiro, Pablo, Chaari, Fakher, Series Editor, Gherardini, Francesco, Series Editor, Ivanov, Vitalii, Series Editor, Haddar, Mohamed, Series Editor, Cavas-Martínez, Francisco, Editorial Board Member, di Mare, Francesca, Editorial Board Member, Kwon, Young W., Editorial Board Member, Trojanowska, Justyna, Editorial Board Member, Xu, Jinyang, Editorial Board Member, Manchado del Val, Cristina, editor, Suffo Pino, Miguel, editor, Miralbes Buil, Ramón, editor, Moreno Sánchez, Daniel, editor, and Moreno Nieto, Daniel, editor

Published
2024
Full Text
View/download PDF

42. Simulation with CAD in Robotics. Free Software Versus Proprietary Software in University Teaching

Author

Tato-Sánchez-Del-Valle, Patricia Eva, Conde-Fernández, Adriana, Vila-Lameiro, Pablo, Chaari, Fakher, Series Editor, Gherardini, Francesco, Series Editor, Ivanov, Vitalii, Series Editor, Haddar, Mohamed, Series Editor, Cavas-Martínez, Francisco, Editorial Board Member, di Mare, Francesca, Editorial Board Member, Kwon, Young W., Editorial Board Member, Trojanowska, Justyna, Editorial Board Member, Xu, Jinyang, Editorial Board Member, Manchado del Val, Cristina, editor, Suffo Pino, Miguel, editor, Miralbes Buil, Ramón, editor, Moreno Sánchez, Daniel, editor, and Moreno Nieto, Daniel, editor

Published
2024
Full Text
View/download PDF

43. Exploring legal age estimation using DNA methylation

Author

Boullón-Cassau, M., Ambroa-Conde, A., Casares de Cal, M.A., Gómez-Tato, A., Mosquera-Miguel, A., Ruiz-Ramírez, J., Cabrejas-Olalla, A., González-Bao, J., Casanova-Adán, L., de la Puente, M., Rodríguez, A., Phillips, C., Lareu, MV., and Freire-Aradas, A.

Published
2025
Full Text
View/download PDF

44. Telomere length and cognitive changes in 7,877 older UK adults of European ancestry

Author

Amy Packer, Leena Habiballa, Esteban Tato-Barcia, Gerome Breen, Helen Brooker, Anne Corbett, Ryan Arathimos, Clive Ballard, Adam Hampshire, Abbie Palmer, Danai Dima, Dag Aarsland, Byron Creese, Margherita Malanchini, and Timothy R. Powell

Subjects
telomere length, cognitive function, polygenic score, ageing, PROTECT study, Geriatrics, RC952-954.6
Abstract

BackgroundTelomere length (TL) has been linked to cognitive function, decline and dementia. This study aimed to explore whether both measured TL and genetic disposition for TL predict dimensions of cognitive performance in a longitudinal sample of older UK adults.MethodsWe analysed data from PROTECT study participants aged ≥50 years without a dementia diagnosis, who had completed longitudinal cognitive testing. We calculated polygenic scores for telomere length (PGS-TL) for 7,877 participants and measured relative telomere length (RTL) in a subgroup of 846 participants using DNA extracted from saliva samples collected within 6 months either side of their baseline cognitive testing. Latent growth models were used to examine whether RTL and PGS-TL predict both baseline and longitudinal changes in cognitive performance (4 time-points, annually).ResultsIn the whole sample, we did not observe significant associations between either measure of telomere length and initial or longitudinal changes in cognitive performance. Stratifying by median age, in older adults (≥ ∼62 years), longer baseline RTL showed a nominal association with poorer baseline verbal reasoning performance (n = 423, Mintercept = 47.58, B = −1.05, p = .011) and PGS-TL was associated with performance over time (n = 3,939; slope factor, Mslope = 3.23, B = −0.45, p = .001; slope2 factor, Mslope2 = 0.21, B = 0.13, p = .002).ConclusionOur findings suggest either the absence of a significant relationship between telomere length (RTL and PGS-TL) and cognitive performance (baseline and change over time), or possibly a weak age-dependent and domain-specific relationship, in older adults of European ancestry. More research is needed in representative and ancestrally diverse samples over a longer assessment period. Alternative biological ageing indicators may still provide utility in the early detection of individuals at risk for cognitive decline (e.g., pace-of ageing epigenetic clocks).

Published
2024
Full Text
View/download PDF

45. Development of the Transfer of Short Story Vehicles to Short Films as a Learning Innovation for Digital Literature Courses

Author

Rianto Rianto and Tato Nuryanto

Subjects
Digital Literature Courses, Short Story Vehicles, Short Films, Education (General), L7-991, Language and Literature
Abstract

This research aims to examine the development of transferring short stories to short films as an innovation in learning Digital Literature courses. The method used is the Research and Development (R&D) model with the ADDIE approach which consists of five stages: analysis, design, development, implementation, and evaluation. At the analysis stage, needs identification is carried out through literature studies, observations, and interviews with lecturers and students to determine current needs and problems in literature learning. The design stage includes material planning, making storyboards and short film scripts, and evaluation design. The development stage includes short film production, the creation of learning materials, initial trials, and revisions based on feedback. Implementation is carried out by involving students in the learning process using short films as a learning medium. The research results show that switching from short stories to short films can increase students' understanding of literary works and provide practical skills in digital media production. This research is expected to offer new insights for teachers and students about the potential for integrating technology in literature learning, as well as motivating the development of more creative and interactive learning methods in the future.

Published
2024
Full Text
View/download PDF

46. Deep learning for understanding multilabel imbalanced Chest X-ray datasets

Author

Liz, Helena, Huertas-Tato, Javier, Sánchez-Montañés, Manuel, Del Ser, Javier, and Camacho, David

Subjects
Electrical Engineering and Systems Science - Image and Video Processing, Computer Science - Computer Vision and Pattern Recognition, Computer Science - Machine Learning
Abstract

Over the last few years, convolutional neural networks (CNNs) have dominated the field of computer vision thanks to their ability to extract features and their outstanding performance in classification problems, for example in the automatic analysis of X-rays. Unfortunately, these neural networks are considered black-box algorithms, i.e. it is impossible to understand how the algorithm has achieved the final result. To apply these algorithms in different fields and test how the methodology works, we need to use eXplainable AI techniques. Most of the work in the medical field focuses on binary or multiclass classification problems. However, in many real-life situations, such as chest X-rays, radiological signs of different diseases can appear at the same time. This gives rise to what is known as "multilabel classification problems". A disadvantage of these tasks is class imbalance, i.e. different labels do not have the same number of samples. The main contribution of this paper is a Deep Learning methodology for imbalanced, multilabel chest X-ray datasets. It establishes a baseline for the currently underutilised PadChest dataset and a new eXplainable AI technique based on heatmaps. This technique also includes probabilities and inter-model matching. The results of our system are promising, especially considering the number of labels used. Furthermore, the heatmaps match the expected areas, i.e. they mark the areas that an expert would use to make the decision.

Published
2022

47. Workshop-based learning and networking: a scalable model for research capacity strengthening in low- and middle-income countries

Author

Perier, Celine, Nasinghe, Emmanuel, Charles, Isabelle, Ssetaba, Leoson Junior, Ahyong, Vida, Bangs, Derek, Beatty, P Robert, Czudnochowski, Nadine, Diallo, Amy, Dugan, Eli, Fabius, Jacqueline M, Baker, Hildy Fong, Gardner, Jackson, Isaacs, Stephen, Joanah, Birungi, Kalantar, Katrina, Kateete, David, Knight, Matt, Krasilnikov, Maria, Krogan, Nevan J, Langelier, Chaz, Lee, Eric, Li, Lucy M, Licht, Daniel, Lien, Katie, Lyons, Zilose, Mboowa, Gerald, Mwebaza, Ivan, Mwesigwa, Savannah, Nalwadda, Geraldine, Nichols, Robert, Penaranda, Maria Elena, Petnic, Sarah, Phelps, Maira, Popper, Stephen J, Rape, Michael, Reingold, Arthur, Robbins, Richard, Rosenberg, Oren S, Savage, David F, Schildhauer, Samuel, Settles, Matthew L, Sserwadda, Ivan, Stanley, Sarah, Tato, Cristina M, Tsitsiklis, Alexandra, Van Dis, Erik, Vanaerschot, Manu, Vinden, Joanna, Cox, Jeffery S, Joloba, Moses L, and Schaletzky, Julia

Subjects
Infectious Diseases, Good Health and Well Being, Partnerships for the Goals, Capacity Building, Developing Countries, Global Health, Humans, Poverty, Students, Universities, Capacity strengthening, Africa, Uganda, research, infectious diseases, Public Health and Health Services
Abstract

Science education and research have the potential to drive profound change in low- and middle-income countries (LMICs) through encouraging innovation, attracting industry, and creating job opportunities. However, in LMICs, research capacity is often limited, and acquisition of funding and access to state-of-the-art technologies is challenging. The Alliance for Global Health and Science (the Alliance) was founded as a partnership between the University of California, Berkeley (USA) and Makerere University (Uganda), with the goal of strengthening Makerere University's capacity for bioscience research. The flagship program of the Alliance partnership is the MU/UCB Biosciences Training Program, an in-country, hands-on workshop model that trains a large number of students from Makerere University in infectious disease and molecular biology research. This approach nucleates training of larger and more diverse groups of students, development of mentoring and bi-directional research partnerships, and support of the local economy. Here, we describe the project, its conception, implementation, challenges, and outcomes of bioscience research workshops. We aim to provide a blueprint for workshop implementation, and create a valuable resource for bioscience research capacity strengthening in LMICs.

Published
2022

48. Empirical treatment against cytomegalovirus and tuberculosis in HIV-infected infants with severe pneumonia: study protocol for a multicenter, open-label randomized controlled clinical trial

Author

Rojo, Pablo, Moraleda, Cinta, Tagarro, Alfredo, Domínguez-Rodríguez, Sara, Castillo, Lola Madrid, Tato, Luis Manuel Prieto, López, Aranzazu Sancho, Manukyan, Lilit, Marcy, Olivier, Leroy, Valeriane, Nardone, Alessandra, Burger, David, Bassat, Quique, Bates, Matthew, Moh, Raoul, Iroh Tam, Pui-Ying, Mvalo, Tisungane, Magallhaes, Justina, Buck, W Chris, Sacarlal, Jahit, Musiime, Victor, Chabala, Chishala, and Mujuru, Hilda Angela

Subjects
Medical Microbiology, Biomedical and Clinical Sciences, Rare Diseases, Pneumonia, Clinical Trials and Supportive Activities, Tuberculosis, Pediatric, Patient Safety, HIV/AIDS, Clinical Research, Pneumonia & Influenza, Lung, Orphan Drug, Infectious Diseases, Evaluation of treatments and therapeutic interventions, 6.1 Pharmaceuticals, Infection, Good Health and Well Being, Child, Clinical Trials, Phase II as Topic, Cytomegalovirus, Cytomegalovirus Infections, HIV Infections, Humans, Infant, Multicenter Studies as Topic, Randomized Controlled Trials as Topic, Treatment Outcome, Valganciclovir, Factorial, Empirical, HIV, Infants, Child mortality, Factorial randomized clinical trial, Cardiorespiratory Medicine and Haematology, Clinical Sciences, Cardiovascular System & Hematology, General & Internal Medicine, Clinical sciences, Epidemiology, Health services and systems
Abstract

BackgroundPneumonia is the primary cause of death among HIV-infected children in Africa, with mortality rates as high as 35-40% in infants hospitalized with severe pneumonia. Bacterial pathogens and Pneumocystis jirovecii are well known causes of pneumonia-related death, but other important causes such as cytomegalovirus (CMV) and tuberculosis (TB) remain under-recognized and undertreated. The immune response elicited by CMV may be associated with the risk of developing TB and TB disease progression, and CMV may accelerate disease caused both by HIV and TB. Minimally invasive autopsies confirm that CMV and TB are unrecognized causes of death in children with HIV. CMV and TB may also co-infect the same child. The aim of this study is to compare the impact on 15-day and 1-year mortality of empirical treatment against TB and CMV plus standard of care (SoC) versus SoC in HIV-infected infants with severe pneumonia.MethodsThis is a Phase II-III, open-label randomized factorial (2 × 2) clinical trial, conducted in six African countries. The trial has four arms. Infants from 28 to 365 days of age HIV-infected and hospitalized with severe pneumonia will be randomized (1:1:1:1) to (i) SoC, (ii) valganciclovir, (iii) TB-T, and (iv) TB-T plus valganciclovir. The primary endpoint of the study is all-cause mortality, focusing on the short-term (up to 15 days) and long-term (up to 1 year) mortality. Secondary endpoints include repeat hospitalization, duration of oxygen therapy during initial admission, severe and notable adverse events, adverse reactions, CMV and TB prevalence at enrolment, TB incidence, CMV viral load reduction, and evaluation of diagnostic tests such as GeneXpert Ultra on fecal and nasopharyngeal aspirate samples and urine TB-LAM.DiscussionGiven the challenges in diagnosing CMV and TB in children and results from previous autopsy studies that show high rates of poly-infection in HIV-infected infants with respiratory disease, this study aims to evaluate a new approach including empirical treatment of CMV and TB for this patient population. The potential downsides of empirical treatment of these conditions include toxicity and medication interactions, which will be evaluated with pharmacokinetics sub-studies.Trial registrationClinicalTrials.gov , NCT03915366, Universal Trial Number U111-1231-4736, Pan African Clinical Trial Registry PACTR201994797961340.

Published
2022

49. Exploring Dimensionality Reduction Techniques in Multilingual Transformers

Author

Huertas-García, Álvaro, Martín, Alejandro, Huertas-Tato, Javier, and Camacho, David

Subjects
Computer Science - Computation and Language
Abstract

Both in scientific literature and in industry,, Semantic and context-aware Natural Language Processing-based solutions have been gaining importance in recent years. The possibilities and performance shown by these models when dealing with complex Language Understanding tasks is unquestionable, from conversational agents to the fight against disinformation in social networks. In addition, considerable attention is also being paid to developing multilingual models to tackle the language bottleneck. The growing need to provide more complex models implementing all these features has been accompanied by an increase in their size, without being conservative in the number of dimensions required. This paper aims to give a comprehensive account of the impact of a wide variety of dimensional reduction techniques on the performance of different state-of-the-art multilingual Siamese Transformers, including unsupervised dimensional reduction techniques such as linear and nonlinear feature extraction, feature selection, and manifold techniques. In order to evaluate the effects of these techniques, we considered the multilingual extended version of Semantic Textual Similarity Benchmark (mSTSb) and two different baseline approaches, one using the pre-trained version of several models and another using their fine-tuned STS version. The results evidence that it is possible to achieve an average reduction in the number of dimensions of $91.58\% \pm 2.59\%$ and $54.65\% \pm 32.20\%$, respectively. This work has also considered the consequences of dimensionality reduction for visualization purposes. The results of this study will significantly contribute to the understanding of how different tuning approaches affect performance on semantic-aware tasks and how dimensional reduction techniques deal with the high-dimensional embeddings computed for the STS task and their potential for highly demanding NLP tasks, Comment: 22 pages, 4 figures and 8 tables

Published
2022

50. BERTuit: Understanding Spanish language in Twitter through a native transformer

Author

Huertas-Tato, Javier, Martin, Alejandro, and Camacho, David

Subjects
Computer Science - Computation and Language, Computer Science - Machine Learning
Abstract

The appearance of complex attention-based language models such as BERT, Roberta or GPT-3 has allowed to address highly complex tasks in a plethora of scenarios. However, when applied to specific domains, these models encounter considerable difficulties. This is the case of Social Networks such as Twitter, an ever-changing stream of information written with informal and complex language, where each message requires careful evaluation to be understood even by humans given the important role that context plays. Addressing tasks in this domain through Natural Language Processing involves severe challenges. When powerful state-of-the-art multilingual language models are applied to this scenario, language specific nuances use to get lost in translation. To face these challenges we present \textbf{BERTuit}, the larger transformer proposed so far for Spanish language, pre-trained on a massive dataset of 230M Spanish tweets using RoBERTa optimization. Our motivation is to provide a powerful resource to better understand Spanish Twitter and to be used on applications focused on this social network, with special emphasis on solutions devoted to tackle the spreading of misinformation in this platform. BERTuit is evaluated on several tasks and compared against M-BERT, XLM-RoBERTa and XLM-T, very competitive multilingual transformers. The utility of our approach is shown with applications, in this case: a zero-shot methodology to visualize groups of hoaxes and profiling authors spreading disinformation. Misinformation spreads wildly on platforms such as Twitter in languages other than English, meaning performance of transformers may suffer when transferred outside English speaking communities., Comment: Support: 1) BBVA FOUNDATION - CIVIC, 2) Spanish Ministry of Science and Innovation - FightDIS (PID2020-117263GB-100) and XAI-Disinfodemics (PLEC2021-007681), 3) Comunidad Autonoma de Madrid - S2018/TCS-4566, 4) European Comission - IBERIFIER (2020-EU-IA-0252), 5) Digital Future Society (Mobile World Capital Barcelona) - DisTrack, 6) UPM - Programa de Excelencia para el Profesorado Universitario

Published
2022

Catalog

Books, media, physical & digital resources
See catalog results