1115254 research outputs found
Sort by
Modeling Electronic Health Records: Interpretable Sequential Approaches for Enhanced Multimodal Embeddings and Future Clinical Outcome Prediction with Time-Aware Irregular Intervals Handling
No full textWith the rapid adoption of Electronic Health Records (EHR), vast amounts of longitudinal and multimodal healthcare data are now available, offering significant opportunities for predictive modeling and clinical decision support. EHR data includes structured elements (e.g., lab results, vital signs) and unstructured clinical notes, capturing a comprehensive patient profile. However, challenges such as irregular time intervals, missing values, and heterogeneous modalities necessitate specialized deep learning models to harness their full potential. This dissertation presents innovative deep learning architectures that enhance the representation, interpretability, and predictive accuracy of multimodal EHR data. To predict Alzheimer’s disease (AD) progression, we developed PPAD, a framework integrating longitudinal and cross-sectional patient data. PPAD employs recurrent neural networks (RNN) and incorporates age at each visit to address time irregularities, significantly improving long-term prediction accuracy. Building on this, we developed TA-RNN, an attention-based time-aware RNN that includes a time embedding layer and a dual-level attention mechanism to enhance interpretability. TA-RNN demonstrated strong performance in predicting AD progression and in-hospital mortality. We further developed CAAT-EHR, a Transformer-based generative embedding model that leverages self- and cross-attention to create task-agnostic longitudinal embeddings. An autoregressive decoder ensures temporal consistency by predicting future visits. Models trained on CAAT-EHR embeddings outperformed those using raw or baseline-processed EHR data across multiple clinical tasks. Collectively, PPAD, TA-RNN, and CAAT-HER, highlight the value of time-aware, interpretable, and generalizable deep learning approaches for improving clinical outcome prediction using multimodal EHR data
Effects of Methylmercury Exposure on Behaviors in Fathead Minnows (Pimephales promelas)
No full textAdverse effects on the neurobehaviors of fishes due to environmental toxicants can have a serious impact. Behaviors such as learning, social behaviors, and aggression are crucial for survival and fitness of the individual and species. Methylmercury is a neurotoxicant that can bioaccumulate in aquatic organisms including fishes and is known to affect several crucial behaviors in various species. The aim of the first study chapter was to design behavioral assays for aggression (mirror test), sociability (sociability assay), learning (T maze assay) and swimming activity (novel tank dive assay) for male and female fathead minnows and characterize control behaviors to see how individual fish perform across multiple behavioral assays and if there were any behavioral correlations across different assays. The behavioral assays from the first study chapter were successfully implemented for fathead minnows with control individual fish performing similarly across different behavioral assays. There were significant correlations of female fish behaviors between the novel tank dive assay and sociability assay. No significant correlations were observed for male fish between the novel tank dive assay and mirror aggression test. The second study chapter aimed to look at the impacts of dietary methylmercury exposure on aggression, social and learning behaviors in adult male and female fathead minnows (Pimephales promelas) using the behavioral assays that were developed in study chapter 1. The results from the second chapter indicated increased aggression in male fish, reduced sociability, reduced swimming activity and reduced learning in female fish. The third study chapter aimed at determining the mercury accumulations and underlying neurotransmitter changes within the tissues post methylmercury exposure in male and female fathead minnows. The results indicated increased mercury accumulations in the muscle, gonads and brain regions of both male and female fathead minnow fish. The female fish accumulated more mercury in their muscle compared to male fish and both sexes accumulated significant mercury in their brain tissues. Neurotransmitters such as dopamine and serotonin concentrations and some of their metabolites were affected in male fish with significant differences across treatments and brain regions which could indicate impairments on the downstream processing and overall metabolism of the neurotransmitters. No significant changes were observed between treatments for female fish. These significant changes in neurotransmitters for male fish further helped to validate the behavioral impairments observed in the study. The results from all the three study chapters demonstrate that methylmercury exposure led to neurobehavioral impairments in both male and female fish and these behavioral impairments were further correlated to the underlying biochemical and physiological changes. This research has both ecological as well as human health relevance and helped in predicting the long-term impacts of methylmercury on fish behaviors
Exploring the Effects of Over-commodification on Heritage Preservation: A Case Study of Cape Coast Castle
No full textThis study focuses on the impact of over-commodification on heritage preservation at Cape Coast Castle, a UNESCO-listed site in Ghana. While heritage tourism presents economic benefits, it also raises concerns about authenticity, community inclusion, and sustainable management. This study adopted a qualitative research design to explore stakeholder perceptions, including locals, site managers, and policymakers, regarding the commodification of the castle. An open-ended interview guide directed data collection, and findings were thematically analyzed. The results of this study indicate significant concerns over economic prioritization and cultural preservation, as well as limited community involvement, commercialization of sacred spaces, and historical narrative shaping at heritage sites. Hence, there is a need for balanced heritage marketing strategies that integrate local participation, ensure cultural preservation, and promotes sustainable tourism
Home at the Crossroads: A Phenomenological Journey of Xicanx Identity in the Borderland through Place, Food, and Language
Full text linkThis dissertation examines the Xicanx community that developed along the Mexican/American border, focusing on the unique identity shaped by the intersection of their cultures. Through a phenomenological examination of my familial experiences, the narrative begins with the journey of my great-grandparents who immigrated from Mexico into America and settled in El Paso, Texas. The physical and metaphysical concept of home is explored in that first chapter tracing the theme of immigration and moving through to my childhood of homelessness. Their establishment of a family in the border-town illustrates their blending of customs between both their Mexican culture and American society. The theme of identity is further explored through food and how Mexican cuisine reflects the Indigenous ancestry of the Chicana/o community. The dissertation also addresses the influence of Nahua and Mayan cultures, including their mythology, traditional meal preparation, and eating practices. The importance of ingredients and culinary tools highlights the deep connection between food and cultural identity within the Mexican community. The dissertation concludes with an investigation into language and the self-designation of the Xicanx people. I argue that the spelling of “Xicanx,” rather than “Chicana/o” offers a more inclusive and accurate representation of the Mexican-American identity. A central theme threaded through this work is the concept of Nepantla, which entails a balance between two cultures. The dissertation hypothesizes that the preservation of culinary traditions, language, and the narratives of our elders serves as a vital means of sustaining their legacy within the community
Methods to Evaluate and Improve the Quality of Text Summarization
Full text linkAutomatic text summarization is a beneficial tool to improve reading efficiency and information accessibility for users. It quickly condenses a large amount of information into key points, helping users to grasp essential information more easily. In addition, it supports various applications such as search engines, recommendation systems, and question-answering systems. However, the quality of automatically generated summaries is sometimes not satisfied. They can be hard to read, omit important contents of the original documents, or even more seriously contain hallucinations. Such issues make the generated summaries less useful and can even mislead the users. Therefore, evaluating the quality of summarization is highly essential. Summarization evaluation is still challenging as there are no ideal evaluation methods. The automatic method is fast, inexpensive, and can handle large volumes of data without human intervention. However, some studies demonstrate that existing automatic evaluation metrics are not well-aligned with human evaluation. Besides, the effectiveness of automatic evaluation metrics has not been thoroughly investigated. Additionally, summarization evaluation is subjective and varies based on users' expectations. As a result, the quality of summarization varies on the target users, indicating that evaluation methods should be customizable and adaptable to meet individual needs. However, conventional automatic evaluation metrics cannot be tailored to individual preferences. While human evaluation can be customized, it is time-consuming and costly. In this thesis, I comprehensively reevaluate existing automatic methods for assessing text summarization, with a focus on abstractive summarization. I also develop new methods to effectively assess the quality of model-generated summaries using LLMs. The proposed evaluation methods are customizable to different quality attributes that users are interested in. Additionally, the developed methods are designed to work in the case when no reference summaries are available. Furthermore, I introduce a new framework to iteratively improve the quality of LLM-based summarization by leveraging the evaluation results. The proposed methods are evaluated on multiple datasets with various domains. Also, I conduct additional analysis to verify the strengths and limitations of the proposed methods
Arquivo.pt Toolkit for Web Archiving [Presentation]
No full textPresentation highlighting the Arquivo.pt toolkit for web archiving which was officially launched in 2023. The Arquivo.pt toolkit for web archiving is an innovative and comprehensive set of services to safeguard digital legacy published online for future generations available to anyone. This presentation will provide a glimpse of these tools and their usage. It was presented at the IIPC General Assembly and Web Archiving Conference held on April 8-10, 2025 in Oslo, Norway
Poking Around in Podcast Preservation
Full text linkPresentation sharing the process for archiving and preserving podcasts from the Netherlands since 2021. Presentation for the IIPC General Assembly and Web Archiving Conference held on April 8-10, 2025 in Oslo, Norway
Automatic Clustering of Domains by Industry for Effective Curation
Full text linkPresentation sharing archive practices in Denmark that takes advantage of existing tools which makes the crawling process more efficient. Presentation for the IIPC General Assembly and Web Archiving Conference held on April 8-10, 2025 in Oslo, Norway
Building a Toolchain for Screen Recording-Based Web Archiving of SVOD Platforms [Presentation]
No full textPresentation sharing the ongoing development of a generic toolchain based on screen recording designed to effectively address DRM restrictions, capture high-quality content, and scale efficiently. It was presented at the IIPC General Assembly and Web Archiving Conference held on April 8-10, 2025 in Oslo, Norway
"What You See No One Saw" [Presentation]
Full text linkPresentation explores a technical and human-centered approach to better preserve the web by focusing on personas—archetypes of web users with distinct behaviors, preferences, and interactions. It was presented at the IIPC General Assembly and Web Archiving Conference held on April 8-10, 2025 in Oslo, Norway