22689 research outputs found
Sort by
The Paradox of Environmental Action: Perceptions of Ecological Decline and Their Association With Well-Being
Get PDFThis dissertation explores a paradox of environmental action: Individuals urgently need to take interest in the plight of our planet’s health, as people’s attitudes, beliefs, and actions will dictate the trajectory of climate change. Yet, evidence shows this kind of engagement increasingly takes a heavy emotional toll on people – so much that they often disengage to protect their well-being. Viewing this paradox as an educational problem positions teaching and learning as crucial solutions to the climate crisis. The paradox is investigated in-depth through three articles.
Article 1 presents a new measurement tool to capture the locus of responsibility for climate action. The survey development process is detailed and evidence of validity is established through implementation in two studies. The first utilized exploratory factor analysis to determine the factor structure of the construct, the second confirmed the survey scale functioned as expected.
Article 2 employs the locus of responsibility scale to examine its associations with emotional outcomes. Regression models are applied in two studies with the outcomes of optimism, negative emotions, and disengagement. Findings from the studies indicate that the locus of responsibility is significantly associated with the outcomes of interest. Additionally, several associations are found between the emotional outcomes and predicting variables of climate concern, nature connectedness, and exposures to climate events. Results of the two studies shed light on the paradox of environmental action by identifying potential contributors to emotional responses and disengagement.
Article 3 utilizes a mixed methods design to examine how undergraduate students navigate the paradox. Specifically, the population is a sub-set of students who face ecological decline in their daily lives as scholars. Key findings include student’s frustrations about the climate crisis and hope for their future, mechanisms to deal with emotions, perceptions of responsibility for mitigating climate change, and reasons to continue fighting for planetary health.
The three articles together investigate the problem of the paradox and offer some solutions for researchers moving forward. Central to this line of inquiry, the perception of responsibility holds promise for those researching emotional responses to the climate crisis and it also advances educational research in this emerging field
BRANCHING MORPHOGENESIS AND COLLECTIVE MIGRATION OF MAMMARY EPITHELIUM DEPEND UPON MICROTUBULE ORGANIZATION AND DYNAMICS
No full textBranching morphogenesis coordinates proliferation and collective cell migration to expand tubular epithelial structures. In the mammary gland, the simple luminal epithelium stratifies into multilayered endbuds to accomplish branching before resolving into simple epithelium. This process requires dramatic changes in cell polarity, proliferation, migration, and actin dynamics. However, little is known about the involvement of microtubules in branching morphogenesis or epithelial migration in 3D environments. Using mammary organotypic cultures, we first examined the organization of the microtubule cytoskeleton during branching. We found that simple luminal cells utilized non-centrosomal apicobasal microtubule arrays, while in stratified luminal cells microtubules radiated from the centrosome. During collective migration, luminal cells adopted an ameboid-like organization with a rear-facing nuclear-centrosomal axis. We also examined microtubule dynamics finding stable microtubule structures in the basal-most luminal cell layer and dynamic microtubules in cells within the stratified layer. Finally, we tested the requirement for microtubule dynamics during branching using microtubule targeting agents. Both microtubule stabilization and destabilization prevented bud formation and arrested duct elongation, independent of effects on mitosis. Cell tracking analysis demonstrated that microtubules coordinated the collective migration of luminal cells in elongating buds. Destabilizing microtubules reduced cell directionality, while stabilizing microtubules maintained directionality but reduced cell motility. Our data reveal that spatiotemporal regulation of microtubule organization and dynamics is essential for collective migration of luminal epithelial cells and mammary branching morphogenesis.
Organoid models recapitulate key features of mammalian tissues and enable high throughput experiments. However, the impact of these experiments may be limited by manual, non-standardized, static, or qualitative phenotypic analysis. OrgDyn is an open-source and modular pipeline to quantify organoid shape dynamics using a combination of feature- and model-based approaches on time series of 2D organoid contours. Our pipeline consists of geometrical and signal processing feature extraction, dimensionality reduction to differentiate dynamical paths, time series clustering to identify coherent groups of organoids, and dynamical modeling using point distribution models to explain temporal shape variation. OrgDyn can characterize, cluster and model unique dynamical paths that define diverse final shapes, thus enabling quantitative analysis of the molecular basis of tissue development and disease
MULTI-COMPARTMENT ASSEMBLOIDS CAPTURE THE ANATOMY AND PHYSIOLOGY OF HEALTHY TISSUES AND TUMORS
No full textIn vitro models are insightful methods used to study the intricacies of cell interactions, molecular targeting, and many other biological phenomena in an environment with greater parameter control and observational capacity than can be achieved with in vivo studies. However, the 3D multicellular and extracellular matrix (ECM) interactions that occur in vivo must be preserved in vitro to maintain physiologically accurate cell behavior. Thus, assembloids, which coculture multiple cell types in 3D to maintain cell-cell and cell-ECM interactions while continuing to facilitate single-cell imaging and molecular resolution, are critically important alternatives to traditional 2D culture and animal models.
This dissertation focuses on multi-compartment assembloids and the significant advances they can contribute to the fields of tissue engineering and cancer biology. I first demonstrate the versatility of the multi-compartment model and the consequences of each parameter modification on the overall architecture of the assembloid. I also demonstrate the model’s ability to mimic complex architectures, such as the intricate folding of the fallopian tube mucosa, in a study that focuses on tissue architecture, function, and molecular expression patterns. Here, I develop a novel platform for validating assembloid models against the tissues they mimic using a combination of computational pathology, comprehensive proteomics, and innovative functional assays. I then apply the multi-compartment model in a new assay aimed at the translation of immunotherapies from the bench to the clinic where an urgent challenge is the infiltration of a solid tumor by engineered immune cells. The multi-compartment assembloid is adapted to isolate each motility and cytotoxicity step of immune cell infiltration to systematically analyze and develop new potential therapies. Finally, I pair a multi-compartment assembloid with a computational model of tumor proliferation and invasion to demonstrate the importance of 3D in vitro systems and the advanced insights that can be obtained from such systems at single-cell resolution.
Multi-compartment assembloids can be customized to capture the anatomy and physiology of healthy tissues and tumors, enabling highly accurate in vitro models. These assembloids are important tools for studies of tissue engineering, tumor progression, drug screening, and many other applications in biology and clinical translation
Interpretable Machine Learning and Deep Learning Frameworks for Predictive Analytics and Biomarker Discovery from Multimodal Imaging Genetics Data
Get PDFIn healthcare, neuroimaging studies and genetics research are generating torrents of data to understand the hereditary components of neurological and psychiatric disorders. Neuroimaging techniques like functional Magnetic Resonance Imaging probe into the neural functioning of the disorder. In parallel, genome sequencing technologies explore the genetic underpinning. Integrating these complementary viewpoints in a single framework improves diagnosis and provides biological insights about the disorders. However, imaging-genetic data lies in a high-dimensional space with complex interaction and unknown causal factors. Our research aims to integrate multimodal imaging-genetics data to predict neuropsychiatric disorders while providing biological insights.
We first propose a novel generative-discriminative framework that integrates imaging and genetics data for simultaneous biomarker identification and disease classification. The generative module extracts representation patterns from the data, while the discriminative module uses the representation vectors for diagnosis. Our experimental analyses show that the discriminative module guides our framework, leading to improved disease diagnosis and biomarker identification. Next, we extend the linear multivariate approach to capture the complex non-linear interaction between the data modalities using an autoencoder framework coupled with a classifier. Unlike traditional encoder-decoder models, our encoder module jointly identifies predictive imaging and genetic biomarkers using Bayesian feature selection.
Our third work uses the Bayesian approach to find genetic variants that causally affect a trait. However, correct identification of the causal variants is challenging due to the correlation structure shared across variants. Our model combines a hierarchical Bayesian model with a deep learning-based inference procedure. We show that this combination provides greater inferential power to handle noise and spurious interactions of the genomic region.
Finally, we solve the problem of encoding whole genome genotype data in an imaging-genetics framework. Traditionally, imaging-genetics models sub-select genetic features to ensure model stability. Our approach departs from conventional Artificial Neural Networks and introduces biologically regularized graph convolution networks to encode the whole genome genotype data. We show that this encoding strategy helps us to track the convergence of genetic risk while preventing overfitting. In an exploratory analysis, we use this model to investigate the underlying biological processes associated with autism spectrum disorder and schizophrenia
The Influence of Surface Roughness and Microstructure on the Fatigue Behavior of Additively Manufactured Ti-6Al-4V
No full textDirect metal laser melting (DMLM) is a method of metal additive manufacturing that has enabled the rapid production of components with complex geometries and increased performance. The adoption of DMLM continues to expand, but there remain additional challenges to overcome. The relatively poor fatigue performance is the result of local defects and stress concentrations inherent to the DMLM process. In this study, the influence of surface roughness and microstructure on the fatigue performance of DMLM Ti-6Al-4V was investigated.
The effect of surface roughness was investigated by performing fatigue experiments to quantify the number of cycles to, and specific location for, crack nucleation. Cross-correlation with micro-CT scans and finite element analysis of the test specimens facilitated identification of critical nucleation sites. While the greatest local stresses arose due to bridging of partially fused powders, these defects did not lead to fatigue failure because they are discontinuous from the bulk of the material. The critical surface features that determine fatigue performance are the deepest valleys. These experimental results were used to create training sets for the development of a deep learning model that can predict the location and intensity of surface stress concentrators in a matter of seconds.
To probe the effect of the underlying microstructure, a new method to study strain localization was developed. The basis of the method is the spatial registration of optical images of surface plasticity captured during a fatigue test with EBSD orientation maps. The results showed that slip steps form at areas of local stress concentration, which arise from an elastic stiffness mismatch between adjacent alpha laths. Transmission electron microscopy revealed that the slip steps form by the activation of prismatic slip, rather than basal. This occurs because the geometric configuration of alpha lath variants that results in a local elastic mismatch also aligns with a high prismatic Schmid factor and a low basal Schmid factor. Since the orientation of an alpha lath and its parent beta grain is governed by a particular Burgers relationship, the fatigue performance of DMLM Ti-6Al-4V may be improved by controlling the crystallographic texture of the prior beta grains
Connectometrics: developing and applying statistical network science towards understanding nanoscale connectomes
Get PDFNeuroscientists have made tremendous progress in our ability to measure the structure of neural circuits. Detailed maps of neural wiring – termed connectomes – are increasingly studied in neuroscience because they have the potential to improve our understanding of how structure relates to function, how the structure of the brain is generated, how it changes with evolution or disease or experience. Despite this progress in measuring connectomes at the scale of individual neurons and the synapses between them, techniques for extracting meaning from these complicated datasets have lagged behind.
This thesis focuses on developing methods for improving our understanding of connectomics data, with a focus on the application of these methods to the connectome of a Drosophila larva brain. In particular, this thesis describes methods for characterizing the general structure of a connectome, including analysis of connectivity-based cell types, characterization of feedforward and feedback structure in a biological neural network, and tools for assessing sensory convergence within the brain using network traversal methods. Then, this thesis presents novel methods for comparing and aligning connectome datasets via a focus on the comparison of the left and right hemispheres of this Drosophila larva brain. The first method enables a statistical comparison of cell type connection probabilities in the left and the right hemispheres, enabling quantitative assessment of which parts of the two connectome datasets have the most significant deviations. The second method enables high-accuracy automated prediction of neuron-to-neuron pairings across the two sides of this connectome by augmenting techniques for graph matching. The tools used and developed as part of this thesis are also made available to the wider neuroscience community and beyond, via the development of documented, tested, open-source Python code to implement these algorithms.
Taken together, this thesis represents an advancement in the algorithmic analysis of connectome data. This kind of analysis will be particularly important going forward, as larger and more complicated connectomes are generated, and in particular, when multiple connectome samples are collected which require quantitative comparison
THE EPIDEMIOLOGY OF DISPARITY AND UTILIZATION TRENDS IN CANCER SCREENING: A LITERATURE REVIEW, RETROSPECTIVE INVESTIGATION, AND COMMERCIAL CLAIMS ANALYSIS IN THE UNITED STATES
No full textObjective: This overarching objective of the dissertation is to conduct a comprehensive examination of the disparities and trends in the utilization of breast, cervical, and colorectal cancer screening among different race/ethnicity groups, insurance groups, and geographical locations.
Methods: This dissertation consists of three distinct yet interrelated studies and aims to: 1) examine racial/ethnic disparities in recent cancer screening uptake in the U.S. via a systematic literature review of relevant studies incorporating atleast one of four large-scale public health survey databases, as provided in the first study; 2) identify changes in the magnitude of cancer screening disparities among race/ethnicity groups in 2008 and 2018, including the effects of insurance, using the National Health Interview Survey (NHIS), as detailed in the second study; and 3) investigate trends in cancer screening utilization among commercially insured adults from 2010 to 2019, focusing on geographical variations, using the MarketScan Commercial Claims database, as explored in the third study.
Results: The findings from the three studies collectively highlighted enduring disparities in cancer screening utilization across select subgroups. The first two studies revealed enduring racial/ethnic disparities in cancer screening utilization, with non-Hispanic (N.H.) Blacks consistently reporting higher odds of receiving cervical and breast cancer screening compared to N.H. Whites (Study I and II), while Asians reported lower odds across cervical (Study I and II). Despite improvements, significant disparities persisted over time, especially for N.H. Asians. The second study demonstrated that the uninsured were disadvantaged across all cancer types. Geographical disparities were also evident, with higher screening rates among urban residents compared with rural residents, as identified in Study 3. Trends in utilization remained stagnant or declined despite the recognized benefits of early cancer detection.
Conclusion: This dissertation highlights the persistent disparities in cancer screening utilization across certain race/ethnicity, insurance, and geographical lines. Despite some improvements, substantial gaps remain, highlighting the importance of addressing underlying factors contributing to disparities among underserved populations and developing targeted interventions
IDENTIFICATION OF SALT-INDUCIBLE KINASES AS NOVEL REGULATORS OF EFFECTOR CYTOKINE PROGRAMS IN LATE-STAGE T EFFECTOR DIFFERENTIATION
Get PDFUnderstanding the mechanisms underlying the acquisition and maintenance of effector function during T cell differentiation is important to unraveling how these processes can be dysregulated in the context of disease and manipulated for therapeutic intervention. We sought to identify the role for the novel mTORC1 substrate, PASK, in T cell effector differentiation. These analyses revealed activation induced upregulation of PASK during T cell differentiation in CD8+ and CD4+ T cells. However, evaluation of PASK function, using the previously reported PASK inhibitor, BioE-1197, revealed PASK independent regulation of T cell effector cytokine programs. Herein, we report the identification of a novel regulator of murine T cell differentiation through the evaluation of this previously unreported activity of the kinase inhibitor, BioE-1197. Specifically, we demonstrate liver kinase B1 (LKB1) mediated activation of salt inducible kinases (SIKs) epigenetically regulates cytokine recall potential in effector T cells. Evaluation of this mechanism revealed late-stage SIK phosphorylation dependent stabilization of histone deacetylase 7 (HDAC7) during effector differentiation. Stabilization increased nuclear HDAC7 levels which correlated with reductions in the activating transcription mark histone 3 lysine 27 acetylation (H3K27Ac) and diminished transcriptional induction of cytokine upon re-stimulation. Inhibition of this pathway, during differentiation, produced effector T cells epigenetically poised for enhanced cytokine recall. This work identifies a novel target for enhancing effector T cell functionality
Exploring a family-level approach to characterize the patterns, predictors, and outcomes of violence against women and adolescents within the home in Ethiopia
No full textIntimate partner violence (IPV) and children (VAC) are global epidemics with shared mental health consequences. While the home serves as a shared locale of exposure and research demonstrates longitudinal linkages between violence exposure in childhood and IPV exposure in adulthood, measurement and intervention strategies commonly use a one-woman or one-child approach, masking overlaps in forms of violence, predictors, and consequences, in the home. To address these gaps, this dissertation drew from socio-ecological, social norms, and family systems theories to address the question, “What are the distinct typologies and predictors of violence against women and their adolescent children in the home, and what is the association between these typologies of violence and adolescent and female caregiver mental health over a one-year period?” A quantitative secondary analysis used two rounds of data from the Gender and Adolescence: Global Evidence cohort study in Ethiopia with 4,195 dyads of adolescents aged 10-12 and their female caregivers. Latent class analysis identified six distinct classes of IPV and three classes of VAC at baseline (Aim 1). IPV classes were characterized by type of violence and included three classes with only emotional and/or economic IPV, and VAC classes were characterized by frequency of violence. Multilevel logistic regression models identified predictors of IPV, VAC, and broad family violence typologies (Aim 2). Risk factors for co-occurring IPV and VAC included caregiver disability, attitudes towards gender inequality and tolerance of VAC, male caregiver alcohol use, women’s sole decision-making, community-level normalization of corporal punishment, and political conflict exposure, alongside protective factors such as positive parenting and higher household economic status. Multilevel linear regression models found that co-occurring IPV and VAC at baseline was associated with greater caregiver and adolescent mental distress one year later (Aim 3). For adolescents, the association was moderated by male caregiver alcohol use and adolescent sex. This study found that a family-level approach reveals distinct typologies of family violence, predictors, and consequences, without losing information specific to women or adolescents. Multisectoral approaches to address men’s alcohol use and normalization of violence may prevent co-occurring IPV and VAC and mental distress among adolescents and female caregivers
Not Working: Transforming Labor in Feminist Art, Film, and Poetry, 1970s-80s
No full textNot Working: Transforming Labor in Feminist Art, Film, and Poetry, 1970s-80s investigates the political and affective economies of housework, maintenance work, clerical work, and craft work in the US through chapter-length studies of feminist art, film, and poetry from the 1970s and 80s. This dissertation brings together the heterogeneous practices of four artists—Chantal Akerman, Faith Ringgold, Mierle Laderman Ukeles, and Karen Brodine—who, while taking markedly different approaches, respond to, represent, reimagine, or reenchant the work that has historically been performed by and associated with women. Through a variety of aesthetic transformations, their art generates alternative ways of thinking about “women’s work” that reject gendered hierarchies of labor and simultaneously suggest novel strategies for resisting work-as-such. Insofar as their artwork breaks with conventions of artistic value—namely that art should exist at a remove from the allegedly banal reproduction of everyday life—the work of these four artists invites us to reexamine still-persistent biases that pervade definitions of avant-garde art. I contextualize my discussions of these artworks in the historical crises of the 1970s and 80s, especially the rise of women’s waged employment, cuts to social spending, financial divestment from cities, and feminist organizing around labor issues