22689 research outputs found
Sort by
3D CHARACTERIZATION OF MICROSTRUCTURAL EFFECT ON SPALL VOID NUCLEATION IN POLYCRYSTALLINE MAGNESIUM
No full textMagnesium (Mg) is a lightweight material with high specific strength, making it suitable for aerospace and automotive applications. This dissertation focuses on studying the behavior of magnesium under shock loading, with a focus on investigating the nucleation of spall voids in polycrystalline magnesium. We present here how microstructural characteristics influence magnesium spall behavior.
Using a laser-driven micro-flyer plate impact setup, we examined the spall response to varying peak shock stresses, strain rates, and grain sizes of three groups of magnesium samples. The spall strength in grain size dependence does not show a constant trend across samples with different initial microstructures, indicating that the batch-to-batch difference may be very critical. We also found that spall strength increases with peak shock stress up to around 3.5 GPa, which may be due to thermal softening. Higher strain rates generally improve spall strength by limiting void growth, although this effect can be offset by shock softening under extreme conditions. The softening mechanism is a reasonable explanation for the experimental observation; however, this conclusion remains tentative due to differences between the batches of samples, including variations in grain size and texture.
The three-dimensional characterization of the orientation mapping of a magnesium sample was done by high-energy diffraction microscopy. The morphology and location of the shock-caused incipient voids were obtained by X-ray tomography. The crystallographic orientations of grains were seeded to generate a statistically equivalent model to compute the normalized Taylor factor across the grain boundaries using crystal plasticity finite element modeling. These voids, typically oblate spheroids, are predominantly located near grain boundaries. Conventional analysis of plastic incompatibility across grain boundaries does not correlate with void nucleation, but grain boundaries with a normalized Taylor factor mismatch greater than 1 are more likely to nucleate voids. On the other hand, the inclination angle of the grain boundary with respect to the loading direction exhibits a strong correlation with the nucleation of the spall void. The observation indicates that shear is also an important factor, as it can resolve the maximum stress and lead to intragranular failure, and grain boundary decohesion is a more dominant factor affecting spall void nucleation
DEVELOPING A MACHINE PERFUSION SYSTEM FOR THE PRESERVATION OF RAT ABDOMINAL WALL VASCULARIZED COMPOSITE ALLOGRAFTS
Get PDFMachine perfusion (MP) systems offer a promising platform for organ preservation, providing continuous nutrient delivery, toxin removal, and treatment administration. This technology has emerged as a vital alternative to static cold storage (SCS), addressing issues related to ischemia-reperfusion injury, high rejection rates, and low recipient allocation. Despite growing interest, MP remains largely experimental, necessitating further development and validation. In this study, we present a systematic approach to designing and manufacturing a custom MP system for small animal vascularized composite allotransplantation (VCA). Our optimized perfusion protocol, utilizing pressure-controlled MP with Perfadex supplemented with components such as colloids, and buffers enabled peripheral perfusion for over 12 hours. This system holds promise for investigating therapeutic delivery to mitigate ischemia-reperfusion injury and enhance transplantation outcomes. Additionally, we explored alternative oxygen carriers, including hemoglobin substitutes and oxygen-generating materials, highlighting their distinct oxygen release patterns and potential applications in MP. Further research is needed to optimize these carriers and improve organ preservation outcomes in VCA MP. Our study lays the groundwork for image-based modeling of vasculature in perfused muscle to investigate oxygen delivery from oxygen carriers
Modulation of brain-wide computation by astrocyte-driven internal states
No full textTo survive, animals must respond appropriately to changing external conditions. To do so quickly and without errors, animals use their past experiences to inform future actions. In response to inescapable stressors, animals initially attempt to extricate themselves from the situation (i.e. active-coping) but might eventually adopt a more passive posture (i.e. passive-coping). Animals learn the futility of their actions and use that to inform future behaviors. However, how futile experience is encoded in the brain and how that modulates sensory processing to change animal behavior is not well understood. In this thesis, I show that larval zebrafish use a representation of their recent futility in astrocytes to dampen the sensitivity of neurons across the brain to visual input and delay motor responses. Larval zebrafish store a representation of their recent futility in the persistent activity of astrocytes, level of norepinephrine and level of dopamine in an environment where swimming does not yield successful translation in space. Elevation of astrocyte activity was sufficient to weaken the sensitivity of neurons to visual motion. These neurons reside in regions involved in transforming visual motion input to motor output. Reduced sensitivity to visual motion across the whole brain likely drive preparatory motor regions more weakly, therefore delaying the optomotor response. Overall, this study shows how the brain leverages astrocytic systems to encode past experience and modulate information processing, which in turn drives animals to switch between active and passive behavioral states
ADVANCING NEXT-GENERATION POLYMERIC PARTICLE ENGINEERING FOR IMMUNOMODULATION AND GENE DELIVERY THERAPEUTICS IN ONCOLOGY AND AUTOIMMUNITY
No full textThe field of immunoengineering is rapidly expanding, fueled by breakthroughs in oncology and autoimmune therapies such as immune checkpoint inhibitors and chimeric antigen receptor (CAR) T-cell treatments. These innovative strategies harness the immune system's power to target and eliminate cancer cells or to regulate its response to autoantigens, opening new avenues for therapeutic intervention. This immunoengineering approaches not only broadens the scope of therapeutic possibilities but also offers targeted, refined interventions that improve patient outcomes by harnessing and modifying the body’s own defenses. The integration of immunoengineering into clinical practice has shown considerable promise, especially through personalized medicine strategies that reeducate the immune system for lifelong adaptation. Research groups globally are now focusing on the creation of innovative particle platforms, each designed to tackle specific medical challenges.
The goal of this thesis is to advance the design of next-generation polymeric particles, aiming to enhance the targeting of disease-relevant cells and the immune system through more efficient drug and nucleic acid delivery. These include improving islet transplants through modulation of T-regulatory cells, reprogramming tumors by delivering mRNA and adjuvants in an antigen-agnostic approach, and promoting immune tolerance via auto-antigen delivering nanoparticles paired with immunomodulatory drugs. Additionally, the work extends to designing vaccines using self-amplifying mRNA and engineering crystallized chemotherapeutic drug formulations for prolonged delivery. Central to this thesis are these transformative strategies that push the boundaries of how biodegradable polymeric particles can be used in medical applications, particularly within the realms of oncology and autoimmunity. The implications of these developments suggest a future where treatment modalities are not only more effective but also more specifically targeted and biocompatible, potentially revolutionizing patient outcomes by addressing diseases at a molecular and cellular level
Iterative Decision Making in Structured Prediction
Get PDFThis thesis investigates structured prediction problems in natural language processing through iterative decision-making, where a model (or agent) makes a sequence of decisions to build one or multiple structured outputs. In Part I, we introduce decision-making in various structured prediction tasks, the mathematical frameworks used to understand these problems, and the machine learning techniques for training such models. We also discuss evaluation metrics, presenting a unified view of various metrics used in prior work and a generalized framework to understand and interpret them, which provides a basis for evaluations in the subsequent parts of the thesis.
Part II focuses on the empirical development of models that adopt the decision-making approach in structured prediction. We investigate a range of semantic structured prediction tasks, from information extraction to semantic parsing. In Chapter 4, we explore a decision strategy that mimics humans’ reading process for event extraction, where an agent iteratively reads text and annotation guidelines to extract elements and build an event structure.
Chapter 5 examines structuring the action space for argument linking, expanding the expressiveness of single actions to predict multiple structural elements simultaneously. This approach is further extended and discussed in Chapter 6 to develop an approach that iteratively predicts multiple structures in a sequence of actions. Chapter 7 shifts the focus from developing specialized models to leveraging large language models for in-context semantic parsing tasks while retaining an iterative decision-making perspective
BELOW COUNTRY: A COLLECTION OF SHORT STORIES
No full textThe South prides itself on being different, and these stories capture the Weird and quirkiness of the region, alongside its distinctive beauty, history and culture. Set in Dakker, a fictional small town along the South Carolina coast, this collection of short stories showcases a wide swath of the town’s people, from the elite to the fringes of society. In each piece, characters wrestle with demons; they struggle against who they are and who they want to be. Some are driven by the desire for money; some for respect and notoriety; and others grasp for a kind of divine ascendency
SPATIAL QUANTITATIVE ANALYSIS OF TUMOR MICROENVIRONMENT REVEALS PREDICTIVE AND PROGNOSTIC BIOMARKERS AT SINGLE-CELL RESOLUTION
No full textCancer is the second most common cause of death worldwide and affects the lives of hundreds of millions of patients. The clinical outcomes of cancer patients are dictated by complex yet highly orchestrated single-cell interactions and contextual cues within tumor microenvironment (TME). Spatial and temporal characterizations of these biological processes are crucial to understanding the mechanisms of tumor control and discovering novel predictive and prognostic biomarkers. In this dissertation, I use the integration of digital pathology, spatial statistics, and artificial intelligence in the context of immuno-oncology to gain insight into the spatial and temporal architecture of TME and its relationship to clinical outcomes, with the overarching goal of identifying novel therapeutic strategies.
To develop the framework, I firstly use high-dimensional tissue imaging to visualize the spatial landscape of tumor microenvironment. Image analysis such as segmentation and classification were applied to decompose TME into spatially resolved single-cell datasets. Next, I apply modern spatial statistics and quantifications to identify features that are associated with clinical outcomes. AI/ML techniques are used to streamline this biomarker discovery process with enhanced robustness, accuracy, and efficiency. Finally, this study makes use of emerging fields that are relevant to our framework, particularly in fields like spatial quantitative systems pharmacology (QSP) and agent-based modeling (ABM), which utilizes the framework output for model parameterizations and calibrations such that the models can simulate the tumor dynamics under various treatments at both spatial and temporal resolutions.
Integrated computational analysis on pathological imaging enables us to decompose the complexity of TME for effective biomarker discovery. The framework identifies patients who are likely to become responders to chemotherapy and immunotherapy in the context of breast cancer, bladder cancer, and liver cancer, and predicts prognosis in patients with pancreatic cancer. Overall, my work aims to provide an effective, robust, and generalizable paradigm to identify clinically relevant signals from complex, unstructured, and high-dimensional imaging data. This research therefore holds promise for optimizing drug dosing regimens, drug effectiveness predictions, and informing clinical trial design
So Close, Yet So Far: Institutional, Partisan, and Theological Factors Inhibiting Women's Representation in United States Politics
Get PDFThis study evaluates whether women representatives in the United States are well-positioned to substantively influence the breadth of policy issues facing state legislatures and the US Congress, why women’s representation has increased more quickly in many areas of the world than in the US Congress, and explores the complementarian religious theologies that discourage formal decision-making authority by women as a contributing factor of the partisan gender gap. The first study compares the committee assignments and chair positions of women representatives in state legislative chambers with high (53.3%-61.2%), median (31.7%-34.7%), and low (11.7%-13.0%) levels of women representatives, finding that parity in women’s representation is not just a symbolic milestone, but increases the breadth of women’s substantive policy influence. At low levels of representation, women are more likely to serve on traditionally women’s issue committees, and some committees have zero women members. It takes between median and high levels of representation before women’s representation on other issue committees begins to catch up with traditionally women’s issue committee membership. Chair positions follow a similar pattern in Democratically controlled legislatures, but not in Republican controlled legislatures where Republican women are typically scarce. The second study charts women’s committee membership and chair positions in the US Congress, finding significant gaps in women’s policy committee membership and chair positions, especially for Republican women; and looks at how the historical context of US institutional development and the emphasis on geographic representation has made it harder for women in the US to gain equal political representation than in many countries whose constitutions were written later. The third study looks at the association between the rise of complementarian religious theologies in the late 1980s and the emergence of the partisan gender gap. Though limited by available data, an initial comparison of Republican primary winners in known high complementarian districts vs other districts suggests a better understanding of, and better data on, complementarian vs. egalitarian church membership could aid strategists in targeting districts with conservative egalitarian evangelicals and less religious conservatives where women are more likely to win GOP primaries
Data-Driven Response: Accelerating Research and Decision-Making in Public Health Emergencies
Get PDFBackground:
This dissertation examines the role of patient and research-related information-sharing practices and the regulatory frameworks that guide them during disaster responses, highlighting their intersection in improving healthcare and public health systems. It addresses the challenges posed by disasters and public health emergencies, intensified by increasing natural disasters and infectious disease outbreaks. By investigating data generation, rapid research protocol implementation, and operational information-sharing barriers, the work provides actionable insights to optimize healthcare and research preparedness and response for future crises.
Methods:
Paper 1: An integrative literature review analyzed English-language articles on rapid emergency human subjects research reviews during disasters, infectious disease emergencies, and public health crises. Articles were selected based on pre-defined criteria, emphasizing empirical research, operational action, or theoretical discussion. Data were synthesized to identify themes, gaps, and recommendations.
Paper 2: The Clinical Characterization Protocol for Severe Emerging Infections (CCPSEI), a prospective observational protocol, was implemented at Johns Hopkins during COVID-19. It streamlined clinical data collection and biorepository development, facilitating a rapid research response during COVID-19 and subsequent high-consequence pathogen outbreaks.
Paper 3: From October 2018 to July 2019, 21 semi-structured interviews with disaster-experienced physicians were conducted using purposive and snowball sampling. Data were analyzed iteratively to identify key themes.
Results:
Robust pre-positioned research protocols, rapid-response ethical frameworks, and improved information-sharing mechanisms are essential for effective disaster and public health emergency response. Paper 1 identified 14 articles, highlighting the importance of administrative preparedness, rigor in rapid reviews, and education/training for IRB members and researchers. Paper 2 enrolled 786 participants between March 2020 and October 2021, demonstrating efficient participant recruitment, minimal clinical care disruption, and expansive downstream research potential through a robust data and biospecimen repository. Paper 3 revealed challenges in patient information management, including the lack of systematic mechanisms for handoffs, limited access to past medical histories, and population-level impacts of information-sharing failures.
Conclusions:
This dissertation emphasizes the need for pre-positioned research protocols, rigorous regulatory frameworks, and efficient information-sharing mechanisms for effective response. Future efforts should focus on implementation and evaluation of frameworks across diverse emergency contexts, informing policy changes to enhance healthcare resilience and research efficacy during crises
An Argument for Keeping Humans in Medicine: The Moral Risks of Giving AI Prescribing Powers
No full textGranting artificial intelligence (AI) autonomous prescribing powers would be a profound mistake that risks undermining the ethical foundations of healthcare. While AI systems offer clear benefits as clinical-support tools, such as assisting in diagnostic imaging, risk prediction, administrative workflows, and treatment recommendations, they lack the compassion, empathy, and ethical reasoning that human clinicians bring to medicine. Policy proposals like the Healthy Technology Act of 2025 (H.R. 238), which would allow AI to prescribe medications, risk devaluing critical human elements of medicine. Compassion and empathy are crucial traits in providing good care, and they are traits that AI systems cannot truly possess, at least for the foreseeable future. Additionally, AI mistakes differ fundamentally from human errors. They are less transparent, harder to trace, and lack moral accountability, posing serious risks to patient trust in the healthcare system. Furthermore, removing human stewardship from prescribing decisions, particularly for antibiotics and opioids, threatens to accelerate public health crises like antimicrobial resistance and opioid addiction. While AI advocates claim that autonomous prescribing could help address healthcare issues such as rural disparities, these issues should instead be addressed through human-centered public health solutions. Ceding clinical decision-making power to AI for short-term gains would set a harmful precedent, making it harder to oppose AI autonomy in even more sensitive areas of medical care. Ultimately, AI should remain a powerful assistive tool under human oversight, not a replacement for human judgment in healthcare decisions