545298 research outputs found
Sort by
Unmodeled DIF in Latent Class Analysis: Consequences for Class Enumeration and Measurement Models
Get PDFThis dissertation builds upon previous research to investigate misspecified covariate effects in latent class analysis (LCA) using Monte Carlo simulations. A key issue in latent class enumeration arises when direct effects (DEs) between covariates and indicators are unmodeled, potentially altering latent class enumeration and the measurement model. Nylund-Gibson & Masyn (2016) recommend that latent class enumeration be completed before incorporating auxiliary variables. However, these recommendations are largely based on simpler models, and whether they hold in more complex conditions remains unclear, particularly when differential item functioning (DIF) is present but unaccounted for. This two-part dissertation builds on prior work by examining more complex modeling conditions, including varying LCA models (i.e., different numbers of classes and indicators), sample sizes, class proportions, DE magnitudes, and multiple covariates. The study evaluates whether unconditional enumeration remains a reliable approach in these complex settings and identifies which fit indices perform consistently well across conditions. It also examines whether latent class solutions remain qualitatively stable when DIF effects are ignored to investigate how class probabilities shift when covariates are introduced and whether DIF can be detected through probability changes rather than enumeration errors alone. Similar to previous research, the results indicate that the Bayesian Information Criterion (BIC) consistently identifies the number of classes, followed by the Consistent Akaike’s Information Criteria (CAIC). Increasing sample size also improves enumeration accuracy, but diminishing effects occur with large sample sizes on misspecified models. Minimal probability shifts in the measurement models observed across replications support the expectation that highly separated and homogeneous classes are resistant to omitted DIF effects. These findings have important implications for the current recommended approach to incorporating auxiliary variables in mixture models
Seismological imaging of the US using complementary datasets reveals mantle processes that shaped the lithosphere
Get PDFThe tectonic plate of the US was constructed from numerous tectonic events, making it an excellent location to study a diverse range of tectonic processes. Models of seismic velocity and anisotropy, the directional dependence of velocity, have proven invaluable for illuminating the interior structure of the Earth. Despite the numerous seismic models of the US, several key features of the lithosphere and asthenosphere remain uncertain. These include the depth of the lithosphere-asthenosphere boundary (LAB), the distribution and character of mid-lithospheric velocity discontinuities (MLDs), seismic anisotropy and its relationship to previous deformation, how the plate causes mantle flow, and how mantle flow influences the evolution of the plate. At the continent-ocean transition of rifted passive margins in particular, little imaging exists to cover the base of the lithosphere or the asthenosphere, making it unclear how continental breakup modified the plate.In this dissertation, I produced seismic models of the lithosphere and asthenosphere of the US. I leveraged the excellent broadband seismometer coverage of the conterminous US along with ocean-bottom seismometers (OBSs) offshore of the eastern North American rifted margin. In Chapter 2, I leveraged this OBS dataset to construct teleseismic shear-wave splitting and differential travel time tomography, producing a velocity and anisotropy model that covers the continent-ocean transition of the mantle. In Chapter 3, I addressed limitations in the well known receiver function H − κ stack method to incorporate radial anisotropy into estimations of the thickness of the crust. In Chapter 4, I conducted a joint inversion of numerous complementary surface wave and receiver function datasets to produce a velocity and anisotropy model of the eastern US. In Chapter 5, I expanded this methodology and data to produce a plate scale model of the conterminous US.While the tectonic events that built the continent likely left a complex signature through the continental lithosphere, features such as sutures are rarely seen through the lithospheric mantle due to limitations in resolution. One of the most striking aspects of my models is the strong correlation of seismic parameters to tectonic boundaries. For example, low wavespeed mantle extends through the lithosphere at the Grenville front, and may be related to an ancient suture. The drop in lithospheric thickness along the Appalachian front appears responsible for the high elevation of the mountain range. I identified sharp changes in crustal thickness and anisotropy that precisely follows the borders of tectonic provinces, including the Basin and Range. Crustal anisotropy tends to be horizontal in extensional regions, even in ancient failed rifts, suggesting the strong influence of crustal deformation on mineral fabrics in the crust.My models suggest that the shape of the lithosphere is a key contributor to mantle flow, causing edge-driven convection and/or shear-driven upwelling. This mantle flow, in turn, erodes and reshapes the lithosphere and continent. Three well-known low velocity anomalies in the mantle of the eastern US are placed near a step in lithospheric thickness, which can be parsimoniously explained as edge-driven convection cells. The asthenosphere offshore of the eastern US appears to actively flow, indicated by anisotropy that is strongly misaligned with plate motion. Despite the tectonic inactivity of the eastern US, these results show that the mantle is highly active. In the western US, I imaged sharp drops in lithospheric thickness along the continental core and Colorado Plateau, which promotes edge-driven convection, contributing to reshaping the western US lithosphere.Our plate-scale models shed new light on the historical and ongoing tectonic evolution of the plate of the conterminous US. Previous tectonic events left a mark on the lithosphere, whose shape controls mantle flow, in turn influencing the evolution of the plate
"What Happened to the VCRs?", or How Legacy Equipment Conservation Became Forgotten within Archival and Library Collections
Get PDFThis thesis aims to demonstrate the importance of legacy playback equipment to accessing analog audiovisual media in libraries and archives. Legacy playback equipment offers users a vital means to experience the content within the media and contextualizes their respective media. These machines should be given just as much attention and care as the media itself. Three principles of traditional art and contemporary time-based media art conservation such as intangible value, significance, and replaceability will be explored as to how they can be applied to the conservation of legacy equipment within libraries and archives. Synthesizing these principles with those of current media archiving will demonstrate how legacy playback equipment cannot be divorced from the audiovisual material. Three case studies underscore the value of legacy equipment in libraries and archives. Each of these case studies will be used to analyze different approaches to the conservation of legacy equipment and raise potential solutions for how to best mitigate the two-front fight of obsolescence and degradation
Challenges and strategies for probing the composite interface of PEM electrolyzers and fuel cells using operando AP-XPS
Get PDFLeft: cross-section schematic of a membrane electrode assembly, the working electrode changing state with applied potential. Center: the operando cell design that enables snapshot data acquisition during trajectory movement. Right: resulting spectra.
Understanding the surface chemistry of electrocatalyst systems under operando conditions is central to revealing the electrocatalytic cell's working mechanisms. Determination of these catalytic processes on a molecular scale and the involved components is fundamental to streamlining material design for energy conversion and storage applications. X-ray photoelectron spectroscopy (XPS) is an established technique used to study the chemical and electronic states of materials. While the surface sensitivity of XPS is typically high, use of tender X-ray energies and technical advancements have allowed for the direct probing of solid–vapor and solid–liquid interfaces. However, protocols and documentation of experimental considerations for operando XPS probing of working electrolyzers and fuel cells remain scarce. Herein, we report an approach for the study of working polymer electrolyte membrane (PEM) electrolysis cells using ambient pressure X-ray photoelectron spectroscopy (AP-XPS). This approach directly probes the composite electrode surface on the membrane electrode assembly (MEA) in 100% relative humidity to establish a meaningful liquid layer for electrocatalysis. We carry out a systematic investigation from the cell constituent components to a fully assembled working operando electrolytic system and establish a method for AP-XPS study of the complex composite MEA, providing recommendations for data acquisition and component analysis
Impact of Metastatic Microenvironment on Physiology and Metabolism of Small Cell Neuroendocrine Prostate Cancer Patient-Derived Xenografts
Get PDFBackground: Potent androgen receptor pathway inhibitors induce small cell neuroendocrine prostate cancer (SCNC), a highly aggressive subtype of metastatic androgen deprivation-resistant prostate cancer (ARPC) with limited treatment options and poor survival rates. Patients with metastases in the liver have a poor prognosis relative to those with bone metastases alone. The mechanisms that underlie the different behavior of ARPC in bone vs. liver may involve factors intrinsic to the tumor cell, tumor microenvironment, and/or systemic factors, and identifying these factors is critical to improved diagnosis and treatment of SCNC. Metabolic reprogramming is a fundamental strategy of tumor cells to colonize and proliferate in microenvironments distinct from the primary site. Understanding the metabolic plasticity of cancer cells may reveal novel approaches to imaging and treating metastases more effectively. Methods: Using magnetic resonance (MR) imaging and spectroscopy, we interrogated the physiological and metabolic characteristics of SCNC patient-derived xenografts (PDXs) propagated in the bone and liver, and used correlative biochemical, immunohistochemical, and transcriptomic measures to understand the biological underpinnings of the observed imaging metrics. Results: We found that the influence of the microenvironment on physiologic measures using MRI was variable among PDXs. However, the MR measure of glycolytic capacity in the liver using hyperpolarized 13C pyruvic acid recapitulated the enzyme activity (lactate dehydrogenase), cofactor (nicotinamide adenine dinucleotide), and stable isotope measures of fractional enrichment of lactate. While in the bone, the congruence of the glycolytic components was lost and potentially weighted by the interaction of cancer cells with osteoclasts/osteoblasts. Conclusion: While there was little impact of microenvironmental factors on metabolism, the physiological measures (cellularity and perfusion) are highly variable and necessitate the use of combined hyperpolarized 13C MRI and multiparametric (anatomic, diffusion-, and perfusion- weighted) 1H MRI to better characterize pre-treatment tumor characteristics, which will be crucial to evaluate treatment response
From Cells to Atlases: Mining Single-cell and Spatial Omics with Graph Learning
Get PDFThe advancements in single-cell and spatial omics technologies have revolutionized our ability to dissect biological systems at unprecedented resolution, allowing us to explore the intricate internal mechanisms of cells, the fundamental units of life, and their complex spatial organization within tissues. However, the high-dimensional, heterogeneous, and often sparse nature of these datasets presents substantial computational challenges. This thesis addresses these challenges by developing a suite of innovative graph learning methodologies designed to systematically analyze single-cell and spatial omics data, spanning a comprehensive range of scales from subcellular genomic organization to multi-slice spatial tissue atlases.We first present our contributions to Genomic and Cellular Analysis. This section introduces scENCORE, a novel method leveraging single-cell epigenetic data and graph embedding to predict chromatin A/B compartment conformations, providing crucial insights into the genome's three-dimensional organization. Then, iHerd is presented as an integrative hierarchical graph representation learning framework, adept at quantifying network changes and prioritizing risk genes in disease contexts. We then introduce ExAD-GNN, an explainable graph neural network that utilizes single-cell data for robust disease state prediction and the identification of cell-type-specific biomarkers, deepening our comprehension of individual cellular complexities.Next, we detail our work on Microenvironmental and Intercellular Understanding within spatial omics. Here, Impeller is developed as a path-based heterogeneous graph learning method, specifically tailored for imputing missing gene expression values within individual cells from spatial transcriptomic data, thus enhancing data completeness. Building on this, iMiracle is introduced as an iterative multi-view graph neural network. This model precisely characterizes intercellular gene regulation from spatial transcriptomic data, elucidating the nuanced mechanisms of cell-to-cell interactions within their native contexts.Finally, our research extends to Large-Scale Spatial Omics Analysis. At the region-level, DISCO offers a diffusion model-based framework. By integrating graph neural networks, DISCO achieves large-scale spatial transcriptomics data completion, effectively addressing the challenges of macroscopic data sparsity. To enable comprehensive joint analysis of multi-slice spatial transcriptomic data and move towards holistic tissue representations, we present MUSE. This computational framework models each spatial slice as a graph and employs optimal transport for robust cross-slice alignment and the construction of integrated spatial atlases.Collectively, the methodologies developed in this thesis underscore the powerful capabilities of graph learning in overcoming the inherent complexities of single-cell and spatial omics data. They offer a cohesive computational paradigm for understanding biological systems across diverse scales, from microscopic genomic features to macroscopic tissue structures and functions, ultimately paving the way for the creation of comprehensive biological atlases
List Decoding Quotient Reed-Muller Codes
No full textReed-Muller codes consist of evaluations of n-variate polynomials over a finite field F with degree at most d. Much like every linear code, Reed-Muller codes can be characterized by constraints, where a codeword is valid if and only if it satisfies all degree-d constraints. For a subset X̃ ⊆ Fn, we introduce the notion of X̃-quotient Reed-Muller code. A function F : X̃ → F is a valid codeword in the quotient code if it satisfies all the constraints of degree-d polynomials lying in X̃. This gives rise to a novel phenomenon: a quotient codeword may have many extensions to original codewords. This weakens the connection between original codewords and quotient codewords which introduces a richer range of behaviors along with substantial new challenges. Our goal is to answer the following question: what properties of X̃ will imply that the quotient code inherits its distance and list-decoding radius from the original code? We address this question using techniques developed by Bhowmick and Lovett [8], identifying key properties of Fn used in their proof and extending them to general subsets X̃ ⊆ Fn. By introducing a new tool, we overcome the novel challenge in analyzing the quotient code that arises from the weak connection between original and quotient codewords. This enables us to apply known results from additive combinatorics and algebraic geometry [34, 35, 37] to show that when X̃ is a high rank variety, X̃-quotient Reed-Muller codes inherit the distance and list-decoding parameters from the original Reed-Muller codes
Non-tenure track faculty in U.S. geography: strategies for support
Get PDFThis article examines the experiences of non-tenure track faculty (NTTF) in geography in U.S. higher educational institutions. While research on NTTF is growing, very little information has been collected in geography. We conducted focus group interviews with 42 NTTF across diverse institutions and found wide variation in roles and working conditions. Participants shared an enjoyment of teaching, but identified challenges related to inclusion, advancement, and workload. They also offered actionable suggestions for departmental and institutional reforms, emphasizing both immediate, practical improvements and longer-term structural changes to support NTTF
A drug repurposing approach reveals targetable epigenetic pathways in Plasmodium vivax hypnozoites
Get PDFRadical cure of Plasmodium vivax malaria must include elimination of quiescent 'hypnozoite' forms in the liver; however, the only FDA-approved treatments are contraindicated in many vulnerable populations. To identify new drugs and drug targets for hypnozoites, we screened the Repurposing, Focused Rescue, and Accelerated Medchem (ReFRAME) library and a collection of epigenetic inhibitors against P. vivax liver stages. From both libraries, we identified inhibitors targeting epigenetics pathways as selectively active against P. vivax and P. cynomolgi hypnozoites. These include DNA methyltransferase inhibitors as well as several inhibitors targeting histone post-translational modifications. Immunofluorescence staining of Plasmodium liver forms showed strong nuclear 5-methylcystosine signal, indicating liver stage parasite DNA is methylated. Using bisulfite sequencing, we mapped genomic DNA methylation in sporozoites, revealing DNA methylation signals in most coding genes. We also demonstrated that methylation level in proximal promoter regions as well as in the first exon of the genes may affect, at least partially, gene expression in P. vivax. The importance of selective inhibitors targeting epigenetic features on hypnozoites was validated using MMV019721, an acetyl-CoA synthetase inhibitor that affects histone acetylation and was previously reported as active against P. falciparum blood stages. In summary, our data indicate that several epigenetic mechanisms are likely modulating hypnozoite formation or persistence and provide an avenue for the discovery and development of improved radical cure antimalarials
Pathologist-like explainable AI for interpretable Gleason grading in prostate cancer
Get PDFThe aggressiveness of prostate cancer is primarily assessed from histopathological data using the Gleason scoring system. Conventional artificial intelligence (AI) approaches can predict Gleason scores, but often lack explainability, which may limit clinical acceptance. Here, we present an alternative, inherently explainable AI that circumvents the need for post-hoc explainability methods. The model was trained on 1,015 tissue microarray core images, annotated with detailed pattern descriptions by 54 international pathologists following standardized guidelines. It uses pathologist-defined terminology and was trained using soft labels to capture data uncertainty. This approach enables robust Gleason pattern segmentation despite high interobserver variability. The model achieved comparable or superior performance to direct Gleason pattern segmentation (Dice score: 0.713±0.003 vs. 0.691±0.010) while providing interpretable outputs. We release this dataset to encourage further research on segmentation in medical tasks with high subjectivity and to deepen insights into pathologists’ reasoning