43569 research outputs found
Sort by
Lexical versus referential meaning in spontaneous speech in psychosis: computational explorations of the semantic space
No full textDisorders of thought involve alterations in linguistic meaning, which can now be approximated using computational models. This thesis targets two dimensions of meaning separately, lexical-conceptual and referential meaning, both of which are affected in schizophrenia spectrum disorder (SSD). The overall aim is to supplement recent computational methods with a more hypothesis-driven approach, and to use this to improve generalizability of findings and deepen our theoretical and methodological understanding in this field. The first study operationalizes referential recurrences in speech as network graphs, with entities manually identified by noun phrases as nodes. This allows referentiality to be studied as a topological problem and reveals distinct distributional patterns in a chronic SSD sample as compared to healthy controls. The second study, conducted within a transdiagnostic sample comprising SDD and major depression, expands the scope of current semantic similarity measures into a dynamic direction suggesting alterations in the geometry of the semantic space. The third study pursues the theme of a ‘shrinking’ semantic space in a new direction, showing through a principal component analysis (PCA) of high-dimensional word embeddings that the semantic space is more reducible in psychosis. In the final study, different measures of semantic similarity are combined to calculate a single overall composite index of how words are associated in speech, which is then demonstrated to track group differences and diagnostic changes over time. Together, these studies advance our understanding of linguistic meaning in psychosis, highlighting both the potential and limitations of analyses based on current computational language models.Els trastorns del pensament impliquen alteracions en el significat lingüístic, cosa que ara es pot aproximar mitjançant models computacionals. Aquesta tesi aborda dues dimensions del significat per separat, el significat lexicoconceptual i el referencial, els quals es veuen afectats en el trastorn de l'espectre esquizofrènic (TEE). L'objectiu general és complementar mètodes computacionals recents amb un enfocament basat en hipòtesis i fer-lo servir per millorar la generalització de les troballes i aprofundir la nostra comprensió teòrica i metodològica en aquest camp. El primer estudi operacionalitza les recurrències referencials a la parla com a gràfics de xarxa, amb les entitats identificades manualment mitjançant frases nominals, com a nodes. Això permet estudiar la referencialitat com a problema topològic i revelar patrons de distribució diferents en una mostra de TEE crònica en comparació amb controls sans. El segon estudi, realitzat en una mostra transdiagnòstica que comprèn TEE i depressió més gran, amplia l'abast de les mesures de similitud semàntica actuals en una direcció dinàmica que suggereix alteracions a la geometria de l'espai semàntic. El tercer estudi aborda el tema d'un espai semàntic cada cop més reduït en una nova direcció, mostrant, mitjançant una anàlisi de components principals (ACP) de word embeddings de gran dimensionalitat, que l'espai semàntic és més reduïble en psicosi. A l'últim estudi, es combinen diferents mesures de similitud semàntica per calcular un índex compost que captura de manera comprensiva com s'associen les paraules a la parla. Aquest índex després demostra ser capaç de rastrejar diferències entre grups i canvis en el diagnòstic al llarg del temps. En conjunt, aquests estudis avancen la nostra comprensió del significat lingüístic a la psicosi, destacant tant el potencial com les limitacions d'anàlisi basades en models de llenguatge computacional actuals.Los trastornos del pensamiento implican alteraciones en el significado lingüístico, lo que puede aproximarse ahora mediante modelos computacionales. Esta tesis aborda dos dimensiones del significado por separado, el significado léxico-conceptual y el referencial, los cuales se ven afectados en el trastorno del espectro esquizofrénico (TEE). El objetivo general de esta tesis es complementar métodos computacionales recientes con un enfoque basado en hipótesis y utilizarlo para mejorar la generalización de los hallazgos y profundizar nuestra comprensión teórica y metodológica en este campo. El primer estudio operacionaliza las recurrencias referenciales en el habla como gráficos de red, con las entidades, identificadas manualmente mediante frases nominales, como nodos. Esto permite estudiar la referencialidad como un problema topológico y revelar patrones de distribución distintos en una muestra de TEE crónica en comparación con controles sanos. El segundo estudio, realizado en una muestra transdiagnóstica que comprende TEE y depresión mayor, amplía el alcance de las medidas de similitud semántica actuales en una dirección dinámica que sugiere alteraciones en la geometría del espacio semántico. El tercer estudio aborda el tema de un espacio semántico cada vez más reducido en una nueva dirección, mostrando, a través de un análisis de componentes principales (ACP) de word embeddings de gran dimensionalidad, que el espacio semántico es más reducible en psicosis. En el último estudio, se combinan distintas medidas de similitud semántica para calcular un índice compuesto que captura de manera comprensiva cómo se asocian las palabras en el habla. Este índice demuestra luego ser capaz de rastrear diferencias entre grupos y cambios en el diagnóstico a lo largo del tiempo. En conjunto, estos estudios avanzan nuestra comprensión del significado lingüístico en la psicosis, destacando tanto el potencial como las limitaciones de análisis basados en modelos de lenguaje computacional actuales.Programa de Doctorat en Traducció i Ciències del Llenguatg
Set covering machine on t-cell receptor LLM representations for lung cancer prediction
No full textTreball fi de màster de: Erasmus Mundus joint Master in Artificial Intelligence (EMAI)Supervisors: Benny Chain and John Shawe-TaylorTutor: Massimo MecellaT-cell receptors (TCRs) provide insights into immune recognition of cancer. We explore whether interpretable rule-based classifiers derived from SCEPTR embeddings of TCR sequences can differentiate cancer repertoires from healthy controls. Using the Set Covering Machine algorithm, we developed models with hyperplane and similarity based rules across alpha and beta chains. Despite strong performance on training data, models failed to generalize to external datasets. Unexpectedly, alpha-chain models often outperformed beta-chain models, and single rules sometimes achieved high training accuracy, suggesting overfitting. Our findings highlight challenges in detecting cancer-specific TCR signatures and indicate current embeddings may capture technical patterns rather than biological signal. We propose future directions including improved rule generation strategies and validation with functionally annotated repertoires
Influencers, gamers y divulgadores culturales: análisis de la cultura de los creadores digitales en España
No full textL’informe present té com a objectiu examinar el fenomen dels creadors digitals en el context espanyol. Per a aquest propòsit, es presenten els resultats de 54 entrevistes realitzades amb tres perfils de creadors: influencers, gamers
i divulgadors culturals. Els resultats s’organitzen en quatre apartats que aborden quatre dimensions claus per comprendre la creació de contingut: les narratives articulades pels participants sobre què significa ser un creador digital; les seves percepcions i relacions amb les plataformes digitals (com ara Instagram, YouTube, TikTok i Twitch); els processos de construcció i manteniment de comunitats de seguidors i de audiències més àmplies; i la naturalesa de les seves interaccions i relacions amb altres creadors digitals. Aquestes dimensions s’analitzen identificant tant les similituds com les diferències entre els tres perfils. A més, l’estudi es veu enriquida per contribucions de tres investigadors convidats que aborden el fenomen dels creadors digitals des de perspectives addicionals: política, en clau de gènere i des del camp editorial. En conjunt, l’informe ofereix una panoràmica completa del paisatge cultural que envolta els creadors digitals i la producció de continguts, amb especial atenció a les seves identitats, pràctiques, impacte i interaccions amb altres actors.Plan de Medidas de Ayuda a la Investigación – Programa COFRE financiado por el Departamento de Comunicación, Facultad de Comunicación, Universitat Pompeu Fabr
Barriers to 12-month treatment of common anxiety, mood, and substance use disorders in the World Mental Health (WMH) surveys
No full textBackground: High unmet need for treatment of mental disorders exists throughout the world. An understanding of barriers to treatment is needed to develop effective programs to address this problem. Methods: Data on barriers were obtained from face-to-face interviews in 22 community surveys across 19 countries (n = 102,812 respondents aged ≥ 18 years, 57.7% female, median age [interquartile range]: 43 [31-57] years; 68.5% weighted average response rate) in the World Mental Health (WMH) surveys. We focus on the n = 5,136 respondents with 12-month DSM-IV anxiety, mood, or substance use disorders with perceived need for treatment. The n = 2,444 such respondents who did not receive treatment were asked about barriers to receiving treatment, whereas the n = 926 respondents who received treatment with a delay were asked about barriers leading to delays. Consistent with previous research, we distinguished five broad classes of barriers: low perceived disorder severity, two types of barriers in the domain of predisposing factors (beliefs/attitudes about treatment ineffectiveness and stigma) and two types in the domain of enabling factors (financial and nonfinancial). Baseline predictors of receiving treatment found in a prior report (i.e., comparing the n = 2,692 respondents who received treatment with the n = 2,444 who did not) were examined as predictors of barriers, while barriers were examined as mediators of associations between these predictors and treatment. Results: Most respondents reported multiple barriers. Barriers among respondents who did not receive treatment included low perceived severity (52.9%), perceived treatment ineffectiveness (44.8%), nonfinancial (40.2%) and financial (32.9%) barriers in the domain of enabling factors, and stigma (20.6%). Barriers causing delays in treatment had a similar rank-order but were reported by higher proportions of respondents (X21 = 3.8-199.8, p = 0.050- < 0.001). Barriers were predicted by low education, disorder type, age, employment status, and financial obstacles. Predictors varied as a function of barrier type. Conclusions: A wide range of barriers to treatment exist among people with mental disorders even after a need for treatment is acknowledged. Most such individuals have multiple barriers. These results have important implications for the design of programs to decrease unmet need for treatment of mental disorders
Computational design and evaluation of peptides to target SARS-CoV-2 spike-ACE2 interaction
No full textThe receptor-binding domain (RBD) of SARS-CoV-2 spike protein is responsible for the recognition of the Angiotensin-Converting Enzyme 2 (ACE2) receptor in human cells and, thus, plays a critical role in viral infection. The therapeutic value of targeting this interaction has been proven by a sizable body of research investigating antibodies, small proteins, aptamers, and peptides. This study presents a novel peptide that impinges the interaction between RBD and ACE2. Starting from a very large pool of structurally designed peptides extracted from our database, PepI-Covid19, a diverse set of peptides were studied using molecular dynamics simulations. Ten of the most promising were chemically synthesized and validated both in vitro and in a cell-based assay. Our results indicate that one of the peptides (PEP10) exhibited the highest disruption of the RBD/ACE2 complex, effectively blocking the binding of two molecules and consequently inhibiting the SARS-CoV-2 spike-mediated cell entry of viruses pseudotyped with the spike of the D614G, Delta, and Omicron variants. PEP10 can potentially serve as a scaffold that can be further optimized for improved affinity and efficacy.This work was supported by Agence Nationale de la Recherche (ANR-RA-COVID-19 (ANR-20-COV4-0001 to RJ)) and La Marato (202130). This work was also supported by grant PID2020-113203RB-I00 and “Unidad de Excelencia María de Maeztu” (ref: CEX2018-000792-M), funded by the MCIN and the AEI (DOI: 10.13039/501100011033), the SGR from the Generalitat de Catalunya (ref: 4413015318- J.SELENT/SGR-22), Biotechnology and Biological Science Research Council (BBS/E/IB/230001D), and the Welsh Government under the Ser Cymru III Tackling COVID program (WG#009). S. A. acknowledges the support provided by Erasmus+ for the study program
Human-AI coevolution
No full textHuman-AI coevolution, defined as a process in which humans and AI algorithms continuously influence each other, increasingly characterises our society, but is understudied in artificial intelligence and complexity science literature. Recommender systems and assistants play a prominent role in human-AI coevolution, as they permeate many facets of daily life and influence human choices through online platforms. The interaction between users and AI results in a potentially endless feedback loop, wherein users' choices generate data to train AI models, which, in turn, shape subsequent user preferences. This human-AI feedback loop has peculiar characteristics compared to traditional human-machine interaction and gives rise to complex and often “unintended” systemic outcomes. This paper introduces human-AI coevolution as the cornerstone for a new field of study at the intersection between AI and complexity science focused on the theoretical, empirical, and mathematical investigation of the human-AI feedback loop. In doing so, we: (i) outline the pros and cons of existing methodologies and highlight shortcomings and potential ways for capturing feedback loop mechanisms; (ii) propose a reflection at the intersection between complexity science, AI and society; (iii) provide real-world examples for different human-AI ecosystems; and (iv) illustrate challenges to the creation of such a field of study, conceptualising them at increasing levels of abstraction, i.e., scientific, legal and socio-political.This work has been partially supported by PNRR - M4C2 - Investimento 1.3, Partenariato Esteso (grant No. PE00000013) - “FAIR - Future Artificial Intelligence Research” - Spoke 1 “Human-centered AI”, funded by the European Commission under the Next Generation EU programme; EU H2020 projects HumaneAI-net G.A. 952026 and SoBigData++ G.A. 871042; European Research Council ERC-2018-ADG 834756 “XAI: Science and technology for the eXplanation of AI decision making”; CHIST-ERA grant CHIST-ERA-19-XAI-010, by MUR (grant No. not available), FWF (grant No. I 5205), EPSRC (grant No. EP/V055712/1), NCN (grant No. 2020/02/Y/ST6/00064), ETAg (grant No. SLTAT21096), BNSF (grant No. KΠ-06-AOO2/5); PNRR (Piano Nazionale di Ripresa e Resilienza) in the context of the research program 20224CZ5X4_PE6_PRIN 2022 “URBAI – Urban Artificial Intelligence” (CUP B53D23012770006), funded by European Union – Next Generation EU; Emanuele Ferragina has been partially supported by a public grant overseen by the French National Research Agency (ANR) as part of the ‘Investissements d'Avenir’ program LIEPP (ANR-11-LABX-0091, ANR-11-IDEX-0005-02) and the Université de Paris IdEx (ANR-18-IDEX-0001). We thank Vincenzo Vivarini for the inspiration about the importance of the feedback loop in complex social systems. There is always a human agent beyond any tactical system. We also thank Daniele Fadda for the support on making the figures. Finally, we thank all members of KDD-Lab for insightful discussions about human-AI coevolution
Mental health, brain resilience and vulnerability in people at risk for Alzheimer's disease
No full textLa enfermedad de Alzheimer (EA) comienza con una fase preclínica donde emergen cambios fisiopatológicos sin síntomas clínicos. Comprender cómo los factores de salud mental se asocian con la vulnerabilidad y resiliencia a la EA es esencial para desarrollar intervenciones efectivas. Esta tesis identifica factores que influyen en las etapas más tempranas cuando las intervenciones pueden ser más efectivas. Los resultados demuestran que eventos estresantes se asocian con la EA, con asociaciones dependientes del tipo de estresor, sexo/género, educación e historia psiquiátrica. Los hallazgos sugieren que síntomas psiquiátricos pueden ser tanto manifestaciones tempranas como contribuyentes a la patología de EA, y que pensamientos intrusivos constituyen una vía psicológica con base neurobiológica en regiones de procesamiento emocional. Las estrategias de afrontamiento adaptativas pueden conferir resiliencia cognitiva. Esta tesis proporciona la primera evidencia que asocia eventos estresantes, síntomas de Trastorno de Estrés Postraumático y estilos de afrontamiento con patología temprana de EA y deterioro cognitivo. Estos hallazgos apoyan intervenciones de salud mental como objetivos terapéuticos para retrasar la enfermedad—especialmente manejo del estrés, tratamiento de síntomas neuropsiquiátricos y mejora de estrategias adaptativas—abriendo nuevas direcciones para investigación sobre este creciente desafío de salud pública.Alzheimer's disease (AD) begins with a prolonged preclinical phase during which pathophysiological changes emerge without clinical symptoms. Understanding how mental health factors associate with vulnerability and resilience to AD during this phase is essential for developing effective interventions. This thesis aims to identify factors that may influence the earliest stages of AD when interventions may be most effective. Results demonstrate that stressful life events associate with AD-related outcomes, with distinct associations contingent on type and timing of stressors, sex/gender, education and psychiatric history. Further, findings suggest that psychiatric symptoms may represent both early manifestations of and contributors to advancing AD pathology. Further, the findings suggest that intrusive thoughts may be a distinct psychological pathway in preclinical AD, having a neurobiological basis in emotion-processing brain regions. Additionally, adaptive coping strategies may confer cognitive resilience against AD pathology. Altogether, this thesis contributes to the growing body of evidence linking psychiatric factors to AD and, importantly, provide the first evidence associating stressful life events, PTSD symptoms and coping styles with early AD pathology and cognitive decline. These findings support mental health interventions as potential therapeutic targets to delay disease trajectories—particularly stress-management, treatment of neuropsychiatric symptoms, and enhancement of adaptive coping strategies in at-risk populations—thereby opening new directions for future research addressing the growing public health challenge of AD.Universitat Pompeu Fabra. Doctorat en Biomedicin
El sistema VioGén i la seva problemàtica judicial
No full textTreball de Fi de Grau en Dret. Curs 2024-2025Tutor: Joan Picó JunoyEl Sistema de Seguiment Integral en els casos de Violència de Gènere (Sistema VioGén) és una eina impulsada pel Ministeri de l’Interior espanyol per avaluar el risc que enfronten les víctimes i coordinar la seva protecció. Aquest treball analitza el seu funcionament i les seves limitacions. Mitjançant l’estudi de la jurisprudència recent dels tribunals espanyol, s’examinen les fortaleses del sistema, com ara la coordinació institucional i l’estimació objectiva del risc de la víctima, i les seves debilitats, com la dependència excessiva del testimoniatge de la víctima i la possible vulneració del principi de presumpció d’innocència, la manca de transparència algorítmica, la valoració poc precisa de les lesions i els possibles biaixos de dades. L’anàlisi conclou que, tot i ser una eina útil i necessària, el sistema encara presenta deficiències estructurals que en limiten l’efectivitat i que poden comprometre no només la protecció de la víctima, sinó també els drets de l’investigat. Amb tot, es proposa una major transparència, millor definició dels indicadors del protocol i una implicació judicial més activa per garantir-ne una aplicació més justa i eficaç
Bridging cells and space: refining Tangram through neighborhood-informed mapping
No full textTreball de fi de grau en Bioinformàtica. Curs 2024-2025Tutors: Richard Rötger, Tomás Bordoy, Lucas Torp DysselSpatial transcriptomics (ST) overcomes the spatial limitations of single-cell RNA sequencing but faces resolution trade-offs. Integrating ST with single-cell (SC) data leverages their complementary strengths, yet current mapping methods like Tangram lack spatial context and exhibit instability. This project analyzes Tangram’s limitations such as mapping inconsistency, bias toward abundant SC reference cell types, and lack of spatial neighborhoods and develops a post-processing framework to enhance biological plausibility. I introduce an Expectation-Maximization (EM) framework that refines Tangram’s probabilistic cell-to-voxel mappings by integrating gene expression similarity and spatial neighborhood information. Evaluated against voxelized MERFISH ground truth, the method improves precision (mean: 0.56) but shows conservative recall (mean: 0.46), highlighting a trade-off between spatial coherence and cell recovery. Our work demonstrates that spatial context refines integration and provides a foundation for more complex neighbor-aware mapping tools
Il caso del porto di Madayi (Kerala, India) e il commercio post-classico nell Oceano Indiano fra tarda Antichità e Medioevo: Una nota preliminare
No full textLe coste sud-occidentali dell'India (il Malabar) hanno rappresentato tradizionalmente un punto di approdo per i commerci e gli scambi fra Medierraneo, Mesopotamia, costa del Mar Rosso, penisola Arabica e Oriente. Le indagini di diagnostica archeologica, la ricognizione di superficie, lo studio delle fonti (cartografia storica ed epigrafi) assieme ad una prima valutazione del dato ceramologico del sito di Madayi, per quanto preliminari, permettono di iniziare a ricostruire una storia che appare essere molto complessa, fatta di relazioni inter-etniche e inter-religiose sullo utilizzando sfondo delle quali si intravedono, tuttavia, le dinamiche economiche che portarono le aristocrazie locali a rafforzare il loro potere sul territorio, utilizzando i proventi ottenuti dallo sfruttamento di queste rotte commerciali