18525 research outputs found
Sort by
The Coevolution of Learned Behavior and Genetic Population Structure
No full textIn order to accurately describe the evolution of traits, researchers must study both their biological and cultural transmission. Socially learned cultural traits have been important for both human and songbird evolution. In humans, social learning influences features such as technology, language, and social structures. Birdsongs are socially learned and have an important role in intra-species competition and mate selection, making birdsong a crucial element in songbird evolution. Culture and genetics evolve together: genetic evolution takes place in an environment made up of cultural traits, and culture can only evolve due to a genetic basis for learning. In this dissertation, I investigate this two-way interaction between genetics and culture in humans and songbirds. I combine computational models with empirical data and simulations to describe the evolution of language, human genetics, and birdsong. I describe how culture has influenced gene flow in human populations, both within and between countries. In human populations, patterns of community organization, kinship, and mobility have all had effects on the geographic distribution of genetic variation, but these effects differed between regions of the world. In birds, I find that the evolution of learning heuristics is affected by the distributions of song traits, and that learning biases have shaped the distribution of songs in a focal species. These analyses of genetic, cultural, and spatial data illustrate the ways in which learned behaviors can shape the dynamics within and between populations
Robust and Flexible Deep Learning Models with Incomplete Medical Image Datasets
No full textDeep learning has revolutionized the field of medical image analysis in the past decade. The success of deep learning can be attributed to the availability of fully annotated medical image datasets. However, the real-world medical image datasets are often heterogeneous and incomplete, such as images with missing modality and partial annotations. Such incomplete datasets significantly hinder the development and deployment of deep learning models for practical use. In this dissertation, several innovative techniques are developed to build robust and flexible deep learning models by exploiting incomplete medical imaging datasets. (1) I propose a generic technique to improve the model robustness to handle incomplete modalities for multi-modality MRI. (2) I explore the feasibility of using synthetic images to diminish the demand of a certain modality for multi-modality imaging. (3) I develop a deep-learning-based landmark localization model for deep brain stimulation with limited human annotation. A comparative study is performed to further evaluate the localization performances of the automatic approaches and the inter and intra-rater variability. (4) To transfer knowledge from one modality to another, I investigate the techniques for unsupervised cross-modality domain adaptation, especially image-level domain alignment. To address the intra-domain variability, I introduce several novel unpaired image translation approaches to improve the style diversity of the translated images. (5) To unleash the synergistic potential of partially labeled datasets, I develop a novel partial label segmentation technique by combining different supervision signals efficiently and effectively. (6) I explore efficient annotation strategies for 3D medical image segmentation by conducting a benchmark study for cold-start active learning approaches
Characterizing Novel Tumor-Microenvironment Interactions in Gli2-Enriched Cancers
No full textWhile cancer treatments have made great strides, aberrant activation of key signaling pathways presents an opportunity for improved therapeutic targeting. Overactivation of the Hedgehog developmental signaling pathway supports tumorigenesis and tumor progression. Using in vitro, in vivo, and computational approaches, the work presented in this dissertation demonstrates novel roles for Hedgehog signaling in modulating how tumor cells interact with their microenvironment in Glioma-associated Oncogene Homolog 2 (Gli2)-enriched cancers. First, we reveal the Hedgehog signaling component Gli2 as a mediator of cytokine expression and secretion by bone metastatic breast cancer cells and show that pharmacologic inhibition of tumor cell Gli2 induces differentiation of myeloid cells toward a pro-inflammatory phenotype. Second, we demonstrate that Gli2 overexpression in dedifferentiated liposarcoma cells induces parallel secretory changes, resulting in a preponderance of M2-like macrophages in an orthotopic tumor model. Additionally, Gli2 overexpression results in a shift toward an osteoblastic gene signature in vitro and phenotypic plasticity of tumor cells in vivo. Finally, this dissertation proposes computational modeling as a strategy to predict experimental outcomes in murine models of bone metastatic cancer, particularly focusing on Gli2-induced processes. In summary, the work presented herein reveals that Hedgehog signaling, and Gli2 more specifically, modifies tumor-microenvironment interactions. We therefore propose Hedgehog inhibition as a strategy for inducing anti-tumor immunity with potential opportunities for combination therapies
Mechanochemical Exploration of the p-Block
No full textAs environmental concerns over organic solvents rise and chemists become more aware of the interference solvents can play in reactions, solvent-free methods have grown in popularity, including mechanochemical methods. Mechanochemical reactions are induced by the direct absorption of mechanical energy, commonly through grinding or milling, with little or no solvent. Organometallic chemistry has advanced through removing the solvent and using mechanochemical methods for synthesis. This dissertation seeks to enhance the fundamental understanding of ball milling through the preparation of bulky allyl complexes of p-block metals, specifically aluminum, germanium, arsenic, and antimony. Heteroleptic aluminum complexes of the formula [(NHC)xAlCl3 nA′n] [NHC = N-heterocyclic carbene; A′ = [1,3-(SiMe3)2C3H3]−; x = 0−1; n = 0−3], were prepared and examined as initiators for L-lactide polymerization, a biodegradable polymer. [AlA′3], which is only accessible via ball milling, was the most active for polymerization. A homoleptic germanium tetra(allyl) ([GeA′4]) was accessed through halide metathesis of germanium halides and K[A′]. [GeA′4] is highly stable, unlike the related [SnA′4], which decomposes when exposed to air or dissolved in solution. The stability of the two systems was explored with computational and experimental studies. Tris(allyl) complexes of arsenic and antimony were prepared with salt metathesis, as two diastereomers; the ratio of diastereomers varies with mechanochemical or solution preparation. The difference in selectivity is attributed to the layered crystal lattices of the EX3 reagents, which template the stereochemistry. The effect of mechanochemical variables on isomer selectivity was explored, including the milling apparatus, milling media, liquid assisted grinding, and use of different reagents. The extent to which the anisotropic lattice is degraded during synthesis affects the diastereomeric ratio. Synthetic chemists commonly find themselves asking the question, what is the best solvent for this reaction? We seek to answer the question: what happens when there is no solvent
Unraveling the Mechanisms of Apical Cadherin-Based Adhesion in Brush Border and Junction Assembly
No full textTransporting epithelia of the kidney and small intestine utilize actin-supported cell surface protrusions, known as microvilli, to expand surface area available for solute transport. Microvilli found on the surface of these epithelia constitute a well-organized “brush border” made up of thousands of protrusions connected via a tip-localized intermicrovillar adhesion complex (IMAC) composed of cadherins CDHR2 and CDHR5. Experiments in this dissertation project revealed that at time points early in differentiation, epithelial cells present two general populations of microvilli: (1) a marginal population at the edges of cells, characterized by high protrusion density, and (2) a medial population characterized by much lower protrusion density. Strikingly, marginal microvilli extend across cell-cell junctions to physically contact microvilli on neighboring cells, using “transjunctional” IMACs. Additionally, transjunctional IMACs are more stable than those bridging medial clusters of microvilli and serve as an anchoring point for nascent microvilli at cell margins. Given the stabilizing nature of transjunctional IMACs, basolateral junctions may be influenced by apical CDHR2/CDHR5 transjunctional contacts. Indeed, in a CDHR2 KO mouse model and in CDHR2 KO CL4 and CACO-2BBE cells, endogenous signal of tight and adherens junction proteins and non-muscle myosin 2-c are reduced. As a result, CDHR2 KO cells exhibit abnormal cell morphologies, increased junction permeability, and impaired collective cell migration. Overall, these findings suggest a new, adhesion-based mechanism for the stabilization of microvilli and support of cell junctions, changing the established understanding of how transporting epithelial cells utilize cell-cell contacts to create optimal tissue structure
Teacher Self-Directed Professional Learning
No full textLeadership and Learning in Organizations capstone projectThis white paper investigates the implementation and impact of a self-directed professional development (PD) model within a school district in Southern California, focusing on experienced teachers’ engagement and growth to meet the evolving educational demands of the 21st century. Utilizing a mixed-methods approach, including surveys, focus groups, and administrator interviews, the study highlights the challenges of aligning traditional PD with veteran teachers’ needs and aspirations. Findings reveal a significant preference among teachers for a more flexible, self-directed approach to PD, underscoring the limitations of a one-size-fits-all strategy. The paper discusses the integration of National Board Certification as a framework for self-directed PD, offering insights into its potential to enhance teacher efficacy, motivation, and student readiness for the future. Key recommendations include allocating dedicated time for PD activities, reevaluating compensation structures to better incentivize participation, and providing targeted training for administrators to support this innovative model. The study contributes to the discourse on teacher professional growth, suggesting that tailored, self-directed PD models can significantly impact instructional quality and outcomes in public education settings
Puente de cuentos para la familia: A community-based, dual language narrative intervention to support language development of young Spanish-English DLLs
No full textThis study aimed to evaluate the impact of a community-based, dual language narrative intervention on young, Spanish-English Dual Language Learners’ (DLLs’) oral language skills. Additionally, this study assessed the extent to which families engaged with the intervention, and the impact of the intervention on children’s home language and literacy experiences. One-hundred parent-child dyads, who self-identified as Hispanic or Latino and spoke Spanish, were randomized to an intervention group and a control group. The 12-week intervention was adapted from a Spanish-English preschool curriculum (Spencer et al., 2020). The Assessment of Story Comprehension, Narrative Listening Measure (English and Spanish), and Puente de Cuentos Vocabulary Assessment (English and Spanish) were administered to children before and after the intervention. Parents also completed a survey to document language use and beliefs pre- and post-intervention. ANCOVA results reveal statistically significant effects for children’s Spanish narrative retell skills, F(1, 97)=19.886, p<0.001. Furthermore, children in the intervention group scored higher on average than children in the control group on measures of English retell skills, story comprehension, and English and Spanish vocabulary, although significant differences were not observed. Parent surveys illustrate the intervention’s positive impact on family’s home language and literacy experiences, parent empowerment, and community building. This study highlights the multifaceted effects of a community-based, dual language intervention aimed at supporting oral language development of young Spanish-English DLLs. Furthermore, findings underscore the impact of early interventions on home literacy and language experiences and the potential for community-based implementation of such programs
The Molecular Basis Underlying Heme Biosynthesis Dysfunction and Disease Originating from C-Terminal Mutations in Aminolevulinic Acid Synthase
No full textAminolevulinic acid synthase (ALAS) is a conserved heme biosynthetic enzyme that catalyzes the first, rate-limiting step of heme biosynthesis in non-plant eukaryotes and alphaproteobacteria. Vertebrates evolved to have two isoforms of ALAS that include the ubiquitously expressed isoform, ALAS1, and the erythroid-specific isoform, ALAS2. ALAS2 takes on the heavy heme requirement that accompanies erythropoiesis. Because of the critical role that ALAS2 plays in developing red blood cells, mutations in ALAS2 lead to functional implications in the heme biosynthesis pathway and result in blood disease. Mutations in ALAS2 lead to a gain-of-function disease, X-linked protoporphyria (XLP), or a loss-of-function disease, X-linked sideroblastic anemia (XLSA). Interestingly, mutations in the ALAS2 C-terminal extension can be implicated in both diseases. The molecular basis for ALAS2 dysfunction mediated by two C-terminal loss-of-function variants, hALAS2 V562A and M567I, was investigated. Two distinct mechanisms of loss-of-function were found to underlie the pathogenesis of XLSA from V562A and M567I. V562A results in decreased enzyme stability and reduced enzyme efficiency concerning succinyl-CoA substrate binding. M567I was found to significantly alter the cooperativity of both glycine and succinyl-CoA substrate binding which coincided with decreased enzyme activity. Together, these data provide the first molecular explanation for loss-of-function underlying XLSA in V562A and M567I. ALAS2 protein interactors were also explored, and it was established that further investigations are needed to dissect the interaction between ALAS2 and its putative binding partner succinyl-CoA synthetase
Patient-specific modeling of cochlear implants
No full textCochlear implants (CIs) use an array of electrodes implanted in the cochlea to directly stimulate the auditory nerve. After CI surgery, patients typically need several appointments with an audiologist over the following months to fine-tune the sound processor's programs for optimal hearing performance. However, few tools exist to assist audiologists in determining the best settings, leading to a time-consuming trial-and-error process that may result in suboptimal outcomes. To address this issue, computational models of the implanted cochlea have been proposed. In this dissertation, we present novel methods for patient-specific modeling of CIs. Firstly, we introduce two model-based applications: auditory nerve fiber health estimation and CI electrode sequence optimization. These proof-of-concept studies provide valuable insights into patient-specific information and have the potential to enhance the hearing outcomes of CI recipients in clinical settings. Secondly, leveraging convolutional neural networks (CNN), we propose novel loss functions and architectures to predict patient-specific electrical parameters and electrical potentials of an electrically evoked cochlea. These CNN models significantly accelerate the modeling process and reduce computational costs for CI users. Thirdly, we develop a transformer-based architecture capable of producing µCT-level tissue label maps of the inner ear using conventional CT scans. This advancement enables more precise modeling of CIs without the need for additional µCT scans. Lastly, we propose a robust and fully-automatic auditory nerve fiber segmentation approach using traditional image-processing techniques, addressing several limitations of a previously used semi-automatic method. In summary, our methods show impressive performance and have made improvements to our current computational models of CIs
Three Essays on Economic Sanctions
No full textEconomic sanctions are a prevalent tool in international relations, used by states to exert influence and achieve foreign policy goals. My dissertation consists of three essays on economic sanction, with particular focus on third-party actors. The first paper, "Who Gets on Board? The Role of Trade Export Similarity in Determining Participation in Economic Sanctions," explores why some third-party states join sanctioning coalitions. Contrary to existing literature that focuses on economic incentives for evading sanctions, I argue that third-party states competing in export markets with the target state may join sanctions to disrupt the target’s trade and gain a competitive advantage. By analyzing commodity-level trade data from 1945 to 2015, I demonstrate that countries with similar export portfolios to the target are more likely to join sanctioning coalitions, particularly those targeting the target's exports.
The second paper, "Shared Trade Partners and the Imposition of Economic Sanctions," examines how the global trade network influences the initiation and success of sanctions threats. I argue that shared trade partners between the sender and the target play a critical role in the sender's ability to exercise economic power. Using network approach, I demonstrate that states are more likely to issue sanctions threats and impose sanctions when they have greater indirect leverage over the target through shared trade partners. I also find that while indirect leverage is positively correlated with the target’s compliance with sanction impositions, it does not significantly impact the target’s response to sanction threats.
The third paper, "Political Risks, Stakeholder Pressure, and Firm Exits: Evidence from the Russian Invasion of Ukraine," shifts focus to the firm-level responses to sanctions. I explore why some multinational corporations (MNCs) comply with sanctions while others continue operations in sanctioned states. I argue that sanctions increase political risks for MNCs in the target state, leading to higher exit rates. However, firms’ political connections and exposure to consumer pressure significantly influence their decisions. Politically connected firms and those facing substantial consumer backlash are more likely to exit. Using data from the Yale School of Management, I estimate how political connections and consumer backlash affect firm exits from Russia.
This dissertation provides theoretical and empirical insights into the motivations behind third-party state participation in sanctions, the impact of global trade networks on sanctions efficacy, and the factors driving firm-level compliance with sanctions. These findings offer valuable perspectives for policymakers and businesses in navigating the complex landscape of international economic coercion