Washington University Medical Center
Washington University St. Louis: Open ScholarshipNot a member yet
26344 research outputs found
Sort by
Three Essays on Methodological Advancements of Large Language Models in Political Science Research
No full textText has become an essential source of data in political science research. Scholars increasingly rely on textual materials—such as legislative speeches, party manifestos, and social media posts—to study the two-way flow of information between elites and the public. Despite this growing reliance, commonly used text-as-data methods often struggle to capture the semantic richness and contextual nuances of political language. Many of these approaches depend heavily on surface-level features like word frequencies and co-occurrence patterns, which constrain their ability to uncover deeper meanings or rhetorical strategies employed by political actors. Meanwhile, advances in computer science have significantly transformed the landscape of natural language processing. A key development was the introduction of pre-trained language models such as BERT, which capture contextual meaning by training on large corpora and fine-tuning for downstream tasks. These models marked a substantial improvement over earlier bag-of-words text representation approaches by enabling more accurate and contextual understanding of texts. Building on this foundation, researchers have continued to scale up model architectures and training data, leading to the emergence of large language models (LLMs) and generative artificial intelligence (AI). These models exhibit extraordinary generalization capabilities, allowing users to perform complex language tasks through plain-language instructions and zero- or few-shot prompt engineering, without requiring extensive labeled data or advanced programming skills. In response, computational social scientists have begun to integrate these new tools into political research. As the scope of available corpora and analytical tasks has expanded, so too have the methodological innovations. Recent work leverages deep neural architectures to extract meaning, builds domain-specific classifiers to label latent features, and applies semantic models to analyze political narratives. This dissertation contributes to this growing body of work by demonstrating how cutting-edge LLMs and AI methods can overcome the limitations of traditional approaches in measuring text similarity and automated frame analysis. Both are important for addressing various questions in areas such as political communication, legislative politics, and democratic representation. I demonstrate the effectiveness of the proposed new methods by applying them to measure media and policy framing, rhetorical polarization, and party message discipline using diverse corpora, spanning from short, informal texts to long, formal documents. This dissertation not only highlights the methodological challenges posed by different types of corpora and analytical tasks, but also develops use cases that illustrate how LLM-based approaches can offer solutions and enable more accurate and nuanced measurement of core political concepts. In Chapter 1, I argue that the most commonly used methods in political science struggle to identify when two texts convey the same meaning as they rely too heavily on identifying words that appear in both documents. This issue is especially salient when the underlying documents are short, an increasingly prevalent form of textual data in modern political research. To address the limitation of current methods, I introduce a transformer model, cross-encoders, which utilizes pair embedding technique that considers the context of both snippets to achieve better estimates of semantic similarity for short texts, such as news headlines and Facebook posts. I illustrate this model in three examples in American politics. First, I apply an off-the-shelf pretrained cross-encoder to measure the similarity between social messages written by experimental subjects and the original Reuters article about the US economic performance that they read in a ``telephone-game\u27\u27 conducted by Carlson (2019), showing that the cross-encoder estimates of information distortion are better at capturing the amount of partisan bias contained in social messages. In the second application which studies the competing media framing of US Supreme Court (SCOTUS), I train a customized cross-encoder model with manually labeled pairs of news headlines to predict the heterogeneity of media coverage of case decisions. The cross-encoder not only outperforms a wide range of word-based and sentence embedding approaches, but also uncover empirical patterns that otherwise would be missed—cases with published dissents receive more diverse coverage than unanimous decisions. The last example presents a more challenging task in which I apply cross-encoder and other models to measure the similarities of social media posts from inter-party and intra-party US senators that a topic model has already identified to be on the same policy issue. Only the cross-encoder yields conclusions that are predicted by established theories, which state that elite polarization is more intensive in domestic policy compared to international affairs. Chapter 2 examines the relationship between institutional power and message discipline using a text similarity approach, which contributes to our understanding of message politics in US Congress. In particular, this chapter implements keyword-assisted topic models for policy agenda coding and OpenAI’s embedding model, which can accommodate extensive context windows of long documents, to generate high-quality text representations for each speech in the congressional record between 1973–2016. This approach allows to estimate the semantic similarity of speech pairs within the same topic, day, and party, serving as a measure of rhetorical unity between leaders and non-leaders. Parties build electorally beneficial brands by staying ``on message.\u27\u27 But when can congressional parties exercise message discipline, who contributes, and how do constituents respond? Building on theories of congressional party discipline, we test a set of competing hypotheses: that institutional power could help or hinder messaging, that the Republican coalition is ideological and homogeneous, and that safe seat members have less constituency concerns. The results show that, generally, institutional power weakens message discipline. However, House Republicans leverage procedural power to offset this disadvantage. Additionally, behavioral evidence suggests that message discipline increases the approval of representatives from copartisan voters. Such results contribute to the literature on message politics and have implications for legislator orientation and thermostatic backlash. Chapter 3 proposes a novel framework of computational frame analysis, which is is central to studies in political communication. Existing methods struggle to identify emerging frames from evolving discourses efficiently. Supervised approaches are labor-intensive and time-consuming. Unsupervised approaches such as topic models and dictionaries rely on clusters of keywords that lack semantic context to capture nuanced framings. Leveraging the extraordinary capability of LLMs in information extraction and summarization, this chapter presents a new chain-of-thoughts prompting method that follows three steps---quote, summarize, and synthesize---to gradually condense concrete text information (e.g., proportions of texts quoted from original documents) into abstract concepts (e.g., frame labels). To discover substantively meaningful frames, I develop a human-AI interaction algorithm to merge similar labels and identify categories. Specifically, I apply GPT-4o to extract emphasis frames in news articles collected by Gilardi, Shipan and Wüest (2021) concerning the adoption of smoking restrictions in public areas across US states. Compared to topic modeling, crowdsourced validation shows that the proposed method produces semantically distinctive and coherent text features for identifying subtly different frames that better align with human interpretation. Moreover, these features reveal frame correlation patterns that align with established theories, which suggests that, as policies diffuse, ideological debates about policy adoption gradually shift toward more technical discussions of policy implementation. Future research can continue to explore the application of LLM-based frame analysis to other policy domains or media narratives. As generative AI continues to advance, integrating such methods into social science research has the potential to significantly improve our capacity to capture the framing of key issues in rapidly changing political discourse
Tank to Bedside: How insights from zebrafish inform our understanding of human melanoma drug resistance
No full textMelanoma, a neural crest-derived cancer, exemplifies the complex interplay of genetic and non-genetic factors driving tumor progression, therapeutic resistance, and phenotypic plasticity. Central to this adaptability is SOX10, a master regulator of neural crest development and melanocyte differentiation, whose expression is controlled by enhancer elements that are reactivated in melanoma. This work provides a comprehensive investigation into the regulatory landscape of SOX10, integrating insights from zebrafish models, human melanoma cell lines, and comparative genomics to uncover the mechanisms underlying SOX10 enhancer activity during development and cancer progression. Using zebrafish as a model system, we identified and validated multiple SOX10 enhancer elements through ATAC-seq, CRISPR-mediated deletions, and enhancer-reporter assays. These studies revealed the spatiotemporal roles of individual enhancers in neural crest development and their reactivation in melanoma. Functional characterization of these enhancers highlighted their critical contributions to SOX10 expression, melanoma cell identity, and the dynamic process of phenotype switching, where melanoma cells transition between drug-sensitive and drug-resistant states. In human melanoma cell lines, deletion of conserved SOX10 enhancer elements led to significant transcriptional and phenotypic shifts, driving cells toward a more invasive, mesenchymal-like state associated with increased drug resistance. RNA-seq and gene set enrichment analyses identified key regulatory networks and candidate genes, that mediate these transitions. Notably, NTRK1 emerged as a promising therapeutic target, with its inhibition restoring drug sensitivity in phenotype-switched, resistant cells. This study further demonstrated the translational potential of cross-species genomic approaches by identifying human SOX10 enhancer elements with conserved functionality in zebrafish. Leveraging CRISPR-based tools, we established a workflow for dissecting enhancer activity and modulating gene expression with precision, offering a scalable method to uncover regulatory mechanisms in development and disease. By elucidating the role of SOX10 enhancers in melanoma plasticity and phenotype switching, this work advances our understanding of the epigenetic regulation of melanoma progression. It provides a foundation for developing novel therapeutic strategies that target enhancer elements to modulate melanoma cell states, combat drug resistance, and improve patient outcomes. This integrative approach underscores the critical importance of regulatory elements in shaping cellular identity, offering new avenues for intervention in melanoma and other neural crest-derived cancers
An Integrated Approach to Western Lowland Gorilla (Gorilla gorilla gorilla) Space Use
No full textSpatial dynamics are physical representations of decisions based on balancing ecological requirements with social dynamics. Socioecological factors are not static, and whether a species can navigate a landscape with greater ecological and social organizational complexity is dependent upon having sufficient abilities for behavioral variability and flexibility. Although the importance and demand for more investigation into intraspecific variation from a species, population, and individual level is recognized, research directly addressing variation is relatively rare in behavioral studies. As such, in this research, we investigated the degree of intraspecific variation in space use behaviors of habituated western lowland gorillas (Gorilla gorilla gorilla) in the Ndoki forest. Specifically, we drew upon a nine-year longitudinal dataset to asses the presence and possible drivers of intraspecific variation in space use behaviors of five habituated western lowland gorilla groups. We investigated the home range size, fidelity, and intergroup overlap, investigating the best predictors of monthly range size, tested whether resource access and dietary profiles were comparable across groups, and evaluated decision-making modalities and influential individuals within given contexts. We found that home ranges were smaller than previous investigations, although they were stable over years. Monthly ranging expanded with fruit availability and group size, but annual differences in range size with seasonality were not pronounced, with annual range as well variable with group size. The greatest degree of variation was with intergroup overlap, which remained consistent over years and seasons. We did not find a difference in gorilla-specific resource access across the groups, with minimal differences in resource access across the home range. The diet profiles of the groups were distinct, but differences in resource access were not the best explanation for differences in diet across groups. Although all individuals demonstrated a high success rate in recruiting followers, females were more successful initiators in the groups. Initiation success similarly differed with context and with the type of initiation. Silverbacks were attempted initiations the most and were generally the most influential individual in group decision-making, but silverback influence was not universal and could change with spatial context and group composition changes. These findings suggest that western lowland gorillas demonstrate a high degree of populational intraspecific variation and flexibility, both in response to ecological and social factors. Further research is necessary to evaluate how space use similarities and differences correspond to differences in resource targeting, whether differences in age and sex class initiation success connects to differences in initiation success with context, and how intergroup social dynamics may connect to differences in ranging dynamics and space use decisions. Based on these findings, I conclude that enhanced cognitive capabilities facilitate behavioral variation and flexibility in western lowland gorillas, and selection for cognitive enhancements that facilitate behavioral variation to multiple socioecological contexts was important within not just human evolution but throughout primate evolution
Poets Against History: American Lyric and Historiography after 1945
No full textThis dissertation argues that American poets after 1945 have competitively engaged with the work of historians in an attempt to transcend history, to correct its gaps, or to change its course. Though literary scholars routinely historicize poetry, and though the last ten years have seen an outpouring of interdisciplinary scholarship on American poetics, we have not considered how the seismic changes in academic historiography after World War II affected the American poets who closely followed those developments. In literary theory, lyric has long been a lightning rod for debates between historicists and formalists because it can signify either a universal poetic impulse as old as human language or a critical fiction that provides aesthetic cover for white supremacy and dubious liberal autonomy. Examining a diverse archive of poets from 1945 until 2020, I offer a counterintuitive thesis: since World War II, historiography has consistently been a whetstone against which poets sharpen their sense of lyric distinctiveness. Poets Against History shows that midcentury confessional poets such as Elizabeth Bishop and John Berryman read deeply in the philosophy of history and Puritan historiography, these intertexts serving as foils to the transcendent aspirations of their lyric art. In the 1970s and ’80s, as American academics increasingly leveraged the tools of social science to write histories from the perspectives of enslaved African Americans and of Native Americans, poets such as Robert Hayden and Joy Harjo used the lyric to imagine voices silenced by the archive and to rehearse timeless myths, defying traditional historical epistemology in the process. For Terrance Hayes, Claudia Rankine, and other contemporary poets, lyric’s taste for solipsism duels with the political imperative to alter the course of history by reckoning with America’s racial past. Ultimately, I argue that to understand the changing principles of American poetry, we must trace changing ideas about American history, and vice versa
Improving metabolomics methods to measure redox balance and understand disease mechanisms
No full textMetabolism serves as the engine of cellular physiology, driving energy production, redox homeostasis and the biosynthesis of macromolecules essential for cell growth and repair. Dysregulated metabolism is increasingly recognized as a hallmark of disease, contributing to the initiation and progression of conditions such as cancer, steatotic liver disease, neurodegeneration and cancer-associated muscle wasting. In this dissertation, I utilize LC-MS-based metabolomics and stable isotope tracing to explore how genetic and environmental perturbations reshape metabolism. In chapter 2, I present a novel, calibration-factor-based LC-MS method for direct and accurate quantification of NAD(P)H/NAD(P)+ ratios, overcoming the limitations of traditional colorimetric assays and eliminating the need for standard curves. In chapter 3, I explore tumor-induced muscle wasting characterized by enhanced BCAA catabolism in a melanoma-bearing zebrafish model and identified ALT as a potential therapeutic target. In chapter 4, I establish NRASQ61R-driven melanoma zebrafish lines to investigate oncogene-specific metabolic alterations and uncover a tumor-liver alanine cycle that helps maintain circulating glucose levels. In chapter 5, I study zinc-dependent metabolic crosstalk between CAFs and PC3 cells, identifying aspartate and asparagine as key metabolites secreted by CAFs that support PC3 proliferation. Together, these studies showcase LC-MS-based metabolomics as a powerful discovery tool to understand how both genetic drivers and environmental factors alter metabolism. The findings in this dissertation provide insights into redox regulation, tumor-host metabolic interactions, and nutrient exchange, offering new avenues for treatments. In addition, this dissertation reinforces the utility of zebrafish as a robust animal model for investigating disease mechanisms, understanding systemic metabolism and identifying potential therapeutic targets
Essays on Macroeconomics
Get PDFThis dissertation is comprised of two essays in macroeconomics. In the first chapter, I theoretically investigate the macroeconomic impact of quantitative easing (tightening) policy on the business cycle. To this end, I build a heterogeneous-agent New Keynesian (HANK) model in which the short-term asset and the long-term asset are imperfect substitutes. In contrast to existing studies, my simulation supposes the QE shock is more persistent in the sense that the central bank announces the future asset purchase strategy in advance. I find that if such an announcement is considered, the power of QE is weaker in the HANK model than in the representative-agent model. Furthermore, excessive extension of QE can have a contractionary effect on the economy. Precautionary saving due to idiosyncratic income risk and redistribution resulting from low long-term yield play a key role in these results. The latter chapter, which is coauthored with Tomiyuki Kitamura, empirically studies the formation of inflation expectations in the firm sector. We construct a small-scale macroeconomic model that incorporates three hypotheses on the formation of inflation expectations: the full information rational expectations (FIRE), inattention, and sticky information hypotheses. Using data for Japan, including survey data on firms\u27 inflation expectations, We estimate the model to examine the empirical validity of each hypothesis, and analyze how incomplete information affects the dynamics of firms\u27 inflation expectations. Our main findings are twofold. First, each one of the three hypotheses has a role to play in explaining the mechanism of the formation of firms\u27 inflation expectations in Japan. Second, although firms\u27 inflation expectations have been pushed up by the Bank of Japan\u27s introduction of its ` price stability target and the expansion in the output gap amid the Bank\u27s Quantitative and Qualitative Monetary Easing (QQE), incomplete information slowed the pace of the rise in firms\u27 inflation expectations
Investigating Molecular Mechanisms of Meiosis and Mitosis in Cryptosporidium parvum
Get PDFThe apicomplexan parasite Cryptosporidium parvum (Cp) is a major global contributor to infectious diarrhea in children and poses a serious threat to immunocompromised adults. This dissertation investigates Cp\u27s unique life cycle, which alternates between sexual and asexual reproduction to achieve three critical objectives: producing oocysts capable of transmission, enhancing genetic variation through recombination, and generating sufficient progeny for survival. Sexual reproduction is essential for forming infectious oocysts that facilitate disease transmission. Yet, the process of meiosis in Cp has been historically understudied due to limitations of traditional in vitro culture methods. The newly developed air-liquid interface (ALI) culture system supports Cp\u27s sexual reproduction in vitro, enabling the exploration of meiosis. Studies using the ALI system and immunocompromised mouse models reveal that Cp chromosomal segregation follows Mendelian inheritance models and displays a high crossover frequency, consistent with observations of Cp\u27s ability to adapt and speciate in nature rapidly. Following excystation, Cp sporozoites invade host cells and rapidly undergo multiple mitotic divisions, which are essential for evading immune responses and establishing infection. Despite the importance of mitosis in Cp pathogenesis, its cellular mechanisms remain unexplored. Using advanced technologies such as CRISPR/Cas9 gene editing, confocal microscopy, and CUT&RUN sequencing, we uncovered features that differentiate Cp from other apicomplexans. We observed a diffuse centromere staining pattern in Cp, distinct from the punctate pattern in Toxoplasma and Plasmodium. Using phospho-histone H3, centrin, and tubulin antibodies, we visualized various stages of Cp mitosis, noting that centromeres remain diffuse even during mitosis. Additionally, Cp telomeres overlap with centromeres at the apical end of the parasite nucleus, rather than at the base as seen in other apicomplexans. We hypothesize that this may be due to the tagged telomere binding protein\u27s occasional occurrence in non-telomeric regions or a tethering of telomeres to the top of the nuclear envelope to be closer to the kinetochore components. This may serve as a mechanism for chromatin organization, epigenetic regulation, etc., and requires additional study. Overall, Cp’s distinctive nuclear architecture may stem from its evolutionary divergence from Toxoplasma and Plasmodium or may be a consequence of its distinct biological adaptations. Cp\u27s unique nuclear configuration may enable the parasite to maintain genome stability while facilitating rapid replication and frequent recombination, overall ensuring the parasite’s adaptability and survival
Two Problems: Hankel Operators and Dyadic Paraproducts
Get PDFParaproducts can be thought of parts of a product of two functions, that isolate particular properties of each of the functions. They have played an essential role in the study of commutators in harmonic analysis, in particular commutators of multiplication by a function and Calder\\u27{o}n-Zygmund operators. In complex analysis, Hankel and Toeplitz operators can be used to decompose a product of two functions. They play the same role as paraproducts in analyzing commutators of certain operators, so they can be thought of as complex analytic analogues of paraproduct operators. The thesis consists of two parts. In the first part, we study a question motivated by the study of Toeplitz operators in real analysis, and classify the boundedness of a composition of two paraproducts. We also establish weighted bounds for certain compositions of paraproducts. In the second part, motivated by what is known about paraproducts in the real valued setting, we consider the question of two-weight boundedness of Hankel operators, with Muckenhoupt weights. We establish conditions under which a Hankel operator is bounded between two weighted spaces, with possibly different weights
Computational and mathematical modeling of microbial community and host responses to microbiota-directed complementary foods administered to children with acute malnutrition.
No full textGlobally, millions of children suffer from undernutrition, with stunting affecting 150 million and wasting nearly 50 million under five years old. Current interventions have shown limited effectiveness in mitigating long-term consequences such as stunted growth, cognitive deficits, and immune dysfunction. Through this body of work, I investigated the efficacy of a microbiota- directed complementary food (MDCF-2) as a next generation therapeutic food to combat childhood stunting. Across two randomized controlled trials involving Bangladeshi children with either moderate acute malnutrition or post-severe acute malnutrition, MDCF-2 significantly outperformed standard ready-to-use supplementary foods (RUSF), leading to significantly improved ponderal and sustained linear growth over extended follow-up periods. Through the integration of plasma proteomic and metagenomic datasets using new computational approaches, I demonstrated that compared to RUSF, MDCF-2 produces an augmented response in the plasma levels of protein mediators and biomarkers of musculoskeletal and central nervous system development and how these mediators are linked to specific growth-promoting members of the microbiota. These findings establish MDCF-2 as a promising therapeutic approach for treating undernourished children and highlight the need to repair the gut microbiome to improve long- term health outcomes in children recovering from varying states of acute malnutrition. Collectively, these discoveries also set the stage for developing point-of-care biomarker panels for improved stratification of populations of children prior to treatment, improved assessment of the efficacy of current therapeutic leads, development of new therapeutic leads, identification of microbial molecular mediators of host responses, and expansion of our understanding of the complicated relationship between the gut microbiome and host systems physiology
Developing Microelectrode Arrays as Multiplex Point-of-Care Diagnostics
Get PDFAntibiotic resistance poses a significant global health challenge, particularly in urinary tract infections (UTIs), where 92% of cases exhibit resistance to at least one antibiotic, causing over 260,000 deaths annually. The emergence of antibiotic-resistant bacteria and growing awareness of the adverse effects of unnecessary antibiotic use highlight the urgent need for better UTI diagnostics. Current state-of-the-art methods are either expensive and time-consuming, or they lack specific details about the nature of the antibiotic and only determine inflammation. This hinders timely and effective treatment decisions regarding whether an antibiotic is actually necessary. Given this backdrop, there is a demand for a complementary method that is affordable, user-friendly, and easily accessible for point-of-care (POC) analysis. Recent developments have shown that prominent uropathogens release specific small molecules during UTIs, which could serve as valuable diagnostic markers. In response to this challenge, microelectrode arrays offer a potential solution, providing a cost-effective, rapid, and label-free approach to monitor multiple binding events and thus provide multiplex sensing of metabolites. This study introduces a diblock copolymer-based microelectrode array platform designed for the quantitative analysis of such small molecules. We demonstrate the ability to site-selectively functionalize this stable, polymer-coated, high-density array with aptamers using the Cu(II)-mediated Chan-Lam coupling reaction. Initial work with borate ester-based polymer surfaces revealed challenges with selectivity due to significant background reactions. These studies revealed significant background fluorescence across the microelectrode array, indicating that the lack of selectivity was not due to a loss in confinement of the desired reaction. Further investigation showed that the background reactions occurred with the reactive borate ester surface even in the absence of the copper reagent used for the Chan-Lam coupling reaction. To overcome this problem, a dual-surface strategy was developed, primarily utilizing a less reactive arylbromide-based polymer that could be selectively converted to the reactive borate ester only at desired electrode sites. This approach significantly enhanced precise aptamer immobilization, reducing background reactions to below 3% and generating stable electrochemical signals for small molecule detection. In simple terms, non-targeted sites on the array lacked borate esters, eliminating borate ester-related background reactions. Further exploration into the reaction mechanism confirmed that the acetylene-functionalized aptamer itself contributes to confinement by forming dimers in the solution above the array, rapidly consuming any Cu(II) reagent that diffuses from the targeted electrodes. Utilizing this optimized platform, three distinct aptamers were successfully placed on a single array that was then used to demonstrate the quantitative and selective electrochemical detection of their respective target metabolites. The polymer-coated microelectrodes were compatible with the multiplex detection of small molecules. Other studies showed that the array-based approach was compatible with the use of human urine samples and demonstrated that the polymer functionalized array was stable for 50 weeks. The array was stable through multiple uses during this “year-long” period, a scenario that clearly illustrated the compatibility of the arrays for a point-of-care application. Taken together, these findings pave the way for constructing robust devices for multiplexed POC detection of metabolites, thereby improving the performance and accuracy of diagnostic tools for the identification of infectious pathogens