22689 research outputs found
Sort by
Educator Influence on the Academic Experiences of Autistic and ADHD Students
This mixed-methods study investigated how general education teachers in kindergarten through fifth grade understand, experience, and implement instructional practices for students with autism and attention deficit hyperactivity disorder (ADHD) in inclusive classroom settings. The study examined how teacher knowledge, self-efficacy, and instructional decisions are shaped by individual beliefs and broader systemic factors such as resource availability, administrative policy, and family–school dynamics. Quantitative data were collected using adapted versions of three instruments: the Autism Self-Efficacy Scale for Teachers, the Knowledge of Attention Deficit Disorders Scale, and the Autism Stigma and Knowledge Questionnaire. Five participants completed all surveys. Qualitative data were gathered through a semi-structured focus group involving three educators. Thematic analysis yielded three core themes: self-efficacy, practices, and knowledge. Findings revealed that teachers expressed strong commitment to inclusive education but reported limited confidence in their knowledge, particularly regarding autism. Teachers reported higher self-efficacy when working with students with ADHD than with students with autism, especially in instruction and social skills support. Although participants demonstrated higher self-efficacy for teaching students with ADHD, they demonstrated greater knowledge about autism. The integration of qualitative and quantitative data highlighted how teachers’ beliefs and instructional decisions were deeply shaped by their training, available supports, and the broader school context, suggesting that improving outcomes for students with autism and ADHD requires enhanced teacher preparation and a reevaluation of systemic supports
Autonomous Flexible Needle Manipulations for Image-Agnostic Percutaneous Needle Interventions
Image-guided percutaneous needle interventions are minimally invasive medical procedures that utilize intraoperative medical imaging to guide needle insertion through the skin for diagnostic and therapeutic purposes. While modern robotic systems can accurately position the needle at the skin entry point, the insertion process remains predominantly surgeon-dependent due to complex needle-tissue interactions that frequently cause path deviations from preoperative plans. Such deviations often necessitate real-time in situ needle adjustments prior to the surgical operation (e.g., biopsy, ablation, or injection), requiring repeated intraoperative imaging and significant clinical expertise.
This thesis introduces a suite of image-agnostic methods to enable fully autonomous needle manipulation during insertion. Adopting a model-driven approach, I first present a mechanics-based needle-tissue interaction model inspired by surgeons' freehand correction techniques. This model is extended into a real-time finite element simulator capable of replicating diverse needle types and manipulation strategies. Leveraging the simulator as a physics engine, I develop novel solutions for needle control and path planning, while demonstrating how image-free feedback modalities can facilitate online model reconstruction for closed-loop control. By bridging simulation with real-world dynamics, this work advances autonomous needle steering systems that reduce reliance on intraoperative imaging and operator skill
ON THE EXCITATORY-INHIBITORY STRUCTURES IN NATURAL AND ARTIFICIAL INTELLIGENCE
Natural intelligences (NIs) display remarkable learning efficiency --- adapting rapidly from limited experience --- whereas artificial intelligences (AIs) often rely on vast datasets and extensive computation. What inductive biases enable such disparity? This dissertation explores the role of biological wiring constraints --- particularly weight polarity --- in driving efficient learning and function representation in both brains and deep networks.
In Chapter 2, we investigate weight polarity, a structural prior shaped during development in natural systems, where synaptic weights retain fixed signs (excitatory or inhibitory) while magnitudes adapt during learning. We show that when polarity patterns are appropriately set a priori, artificial networks learn faster with fewer samples. However, we also delineate scenarios where fixing polarity casts a disadvantage.
In Chapter 3, we address the question: why do brains and deep networks have negative (inhibitory) weights? We leverage the universal approximator theorem and prove that networks with non-decreasing activation functions and non-negative weights are not universal approximators. This result provides the first general justification --- beyond function-specific explanations --- for the necessity of inhibitory connections in brains and negative weights in artificial networks, offering geometric insights into the functional limitations of purely excitatory architectures.
To further bridge structure with function, in Chapter 4, we develop XORness, a normative and testable measure of functional complexity. Using whole-brain electron microscopy (EM) connectomes of larval and adult Drosophila, we show that actual brain networks --- and synthetic networks with matched connectivity statistics --- exhibit high XORness, suggesting that evolution favors topologies capable of representing complex nonlinear functions. Specifically, our simulations predict that maximal functional complexity occurs at 72% and 79% excitatory neurons in larval and adult Drosophila, closely matching empirical estimates (67% and 74%, respectively). These optima also require higher connectivity in inhibitory neurons --- another feature corroborated by EM data.
Together, this work reveals how biological constraints on polarity shape the functional capacity of both natural and artificial intelligences
Coercion, Control, And Community: Ethnic Contexts and Public Health Implications of Intimate Partner Violence in Suriname
BACKGROUND: Intimate Partner Violence (IPV) remains a pervasive public health issue globally, with significant physical, psychological, and socio-economic consequences. In Suriname, a pluralistic society characterized by distinct ethnic groups and sociocultural norms, IPV prevalence exhibits notable variation across ethnic lines. Despite existing research, there is a paucity of detailed, context-sensitive analysis of IPV correlates, particularly those associated with ethnicity and community-level factors. This research was guided by the Social Ecological Model and complementary theoretical frameworks, providing a comprehensive lens for analyzing IPV's multifaceted determinants.
OBJECTIVES: This dissertation examined the sociocultural factors that shape women’s experiences of IPV in Suriname. The aims were to: (1) Synthesize the state of the research on community-level risk and protective factors in low- and middle-income countries, based on studies published since 2011 and utilizing nationally representative datasets. (2) Describe differences in prevalence of lifetime IPV, gender and IPV attitudes, and social capital among ever-partnered women from the five major ethnic groups in Suriname and assess whether variation in lifetime IPV across ethnic groups persisted after controlling for explanatory variables. (3) Characterize patterns of IPV, examine if prevalence of IPV patterns differ by ethnicity, and assess whether variation in patterns of IPV across ethnic groups persists after controlling for explanatory variables.
METHODS: Guided by the socio-ecological model for IPV, this research employs a multi-method quantitative approach, comprising of a scoping literature review of community-level risk and protective factors for IPV, Pearson F tests, regression analyses, and latent class analysis of data from the 2018 Suriname Women’s Health Survey.
RESULTS: The scoping review highlights the significance of community-level factors, including levels of IPV-accepting attitudes, exposure to violence, and young age at marriage, while the level of women’s autonomy was protective. Community wealth, employment, and education served as risk factors in some settings and protective factors in others. Analysis of Surinamese data reveals substantial variations in IPV prevalence and associated factors across ethnic groups, with social capital and partner alcohol misuse emerging as key correlates. Latent class analysis identified three distinct patterns of IPV experiences: (1) Low Violence/Moderate Controlling Tactics, (2) Moderate Violence & Moderate Controlling Tactics, and (3) Systematic Abuse.
CONCLUSIONS: The findings underscore the importance of considering community and cultural contexts in IPV research and interventions. Tailored, culturally sensitive strategies are essential for addressing IPV in Suriname and other pluralistic societies. The dissertation contributes to the broader understanding of IPV determinants, emphasizing the need for culturally informed interventions, community engagement, and policy reforms to effectively address IPV and support survivors. For nursing practice in particular, the results highlight the value of (a) screening instruments that capture the full spectrum of IPV typologies, (b) trauma- and culture-informed communication that validates survivors’ diverse experiences, and (c) interdisciplinary collaboration to develop personalized safety-planning and referral pathways, thereby positioning nurses as pivotal actors in both the prevention of IPV and the provision of survivor-centered care
BRIDGE THE GAP BETWEEN ORGANIC SYNTHESIS AND BIOSYNTHESIS: REPROGRAMMING METALLOENZYMES FOR NEW-TO-NATURE REACTIVITIES
As the catalysts of nature, enzymes offer unmatched selectivity, efficiency, and sustainability, making them increasingly attractive for modern chemical synthesis. Recent advances in protein engineering techniques and mechanistic understanding have paved the way of reprogramming enzymes to catalyze reactions far beyond the chemistries in nature. This thesis focuses on the development of novel biocatalytic platforms using non-heme iron enzymes, a versatile class of metalloenzymes traditionally involved in radical-mediated hydroxylation and halogenation. Through a combination of metal center substitution, first-sphere residue engineering and directed evolution, these works expand the functional landscape of non-heme enzymes to access diverse non-natural reactivities.
In the first part of this thesis, a non-heme iron enzyme SadA was successfully reprogrammed to catalyze the Conia-ene reaction, a Lewis acid-mediated transformation widely applied in organic synthesis but previously unknown in enzymology. Through rational substitution of the native iron center with Cu(II) and subsequent directed evolution, an optimized variant was obtained with excellent yield, enantioselectivity, and catalytic efficiency. This represents the first example of biocatalytic Lewis acid mediated ene reaction, establishing a new framework for bridging enzymatic and organometallic catalysis. The second part of the thesis addresses the challenge of biocatalytic enantiodivergence, i.e. accessing both enantiomers of a product using related enzyme scaffolds. By rationally mutating the first coordination sphere of a non-heme iron enzyme HppE and substituting its metal center to copper, two complementary variants were developed offering opposite enantiomers of Conia-ene products. This work introduces a general strategy for achieving enantiodivergence in biocatalytic systems. In the third part, a non-natural N-radical hydroamination reaction was developed using a reprogrammed non-heme iron enzyme. Inspired by non-heme enzymes’ ability to activate N-F bonds and generate nitrogen-centered radicals, a tailored substrate was designed to undergo intramolecular radical cyclization to form valuable γ-lactam scaffolds catalyzed by a reprogrammed non-heme iron enzyme PAH. We anticipate these findings illustrated in this thesis would not only deepen our understandings of reprogramming metalloenzymes for non-natural functions, but also lay a foundation for the future industrial application of biocatalysis in the development and manufacturing of complex, pharmaceutically relevant molecules
Chemotherapy-induced endocycling cancer cells have altered chromatin organization and functional transcriptional changes
Therapy resistance continues to pose a significant barrier in effectively treating advanced cancers. Although initial responses to treatment can result in the clinical response of a reduction in tumor size, emergence of resistant cell populations results in disease recurrence and progression. Recent research suggests that resistance involves cellular plasticity beyond genetic mutations alone, implicating chromatin remodeling and transcriptional reprogramming as key contributors to cancer cell survival after chemotherapy exposure. This thesis explores the response of cancer cells to chemotherapy, focusing on changes in nuclear structure, chromatin accessibility, and gene expression profiles. In prostate (PC3) and breast (MDA-MB-231) cancer cell line models treated with cisplatin, we identified a subset of cells that survive chemotherapy and become polyploid through accession of an endocycling cell cycle program. We observed increased nuclear size, chromatin decompaction, and dysregulation of key chromatin-modifying enzymes in these cisplatin-resistant, surviving cells. Bulk ATAC-sequencing revealed significant alterations in chromatin accessibility at promoters and enhancers, with a notable shift from open chromatin at proliferative genes to genes implicated in inflammatory and stress-response pathways. By integrating the chromatin accessibility data with RNA-sequencing data, we confirmed substantial transcriptional reprogramming that further emphasized the activation of inflammatory signaling pathways and reduced cell cycle progression. Collectively, these findings demonstrate that chromatin remodeling and transcriptional changes drive chemotherapy-induced cellular plasticity in a non-proliferative, polyploid population of cisplatin-resistant cancer cells, offering potential targets for new therapeutic strategies against treatment-resistant cancer populations
RADical Shifts: A Futurist's Guide to Ecological Transformation and Biodiversity Stewardship
RADical Shifts: A Futurist’s Guide to Ecological Transformation and Biodiversity Stewardship provides natural resource managers and conservationists with practical tools to navigate the unprecedented challenges posed by climate change. The guide introduces the Resist—Accept—Direct (RAD) framework, offering a flexible, adaptive approach to managing ecological transformation. By using RAD, conservation leaders can identify strategies to resist, accept, or direct ecological changes in ways that enhance biodiversity stewardship.
The guide emphasizes that climate change can be meaningfully addressed through informed planning and on-the-ground action. It introduces the concepts of RAD menus, RAD portfolios, and RAD decision context, which help managers brainstorm adaptation strategies, track decisions over time and space, and adapt decision-making processes.
Successful RAD actions are often the culmination of many years of collective work, not easily visible from an outsider’s perspective. Radical Shifts helps to demystify these processes. It includes case studies that examine the behind-the-scenes realities of RAD decisions in different regions of the United States.
With RAD menus, managers can explore a full spectrum of adaptation actions. RAD portfolios assist in planning and tracking these decisions, accounting for both spatial and temporal factors. The guide also stresses the importance of collaborative, deliberative engagement to adjust social and institutional contexts in response to changing ecological conditions. It notes that failure to adapt decision-making processes can obstruct progress, as past values, rules, and knowledge may hinder the ability to respond to change.
Whether managing a small-scale project or leading larger efforts, this guidebook equips natural resource managers with adaptable approaches to manage biodiversity and ecosystem services in an era of ecological uncertainty. It empowers conservation leaders to act now while embracing the ongoing journey of learning and adaptation, making it an essential resource for navigating the complexities of climate change and ecological transformation
INTEGRATING COMPUTATIONAL CHEMISTRY AND MACHINE LEARNING FOR NEXT-GENERATION MATERIALS DISCOVERY
Computational chemistry can draw upon a plethora of algorithms and techniques within its ``playbook'' that can help decipher the properties of materials and processes. Beyond enhancing our understanding of well-established materials, it can be used to provide valuable insights into novel, less-explored materials, or even propose new ones that can sometimes escape experimental scrutiny. In this work, we focus on how to approach computational techniques and efficiently integrating them into the material discovery pipeline.
Since a key part of chemistry is built upon understanding transitions from an initial to a final state along an often unknown reaction coordinate, we help to build a better understanding of a key method that exemplifies building such reaction coordinates, the well-established Nudged Elastic Band (NEB) algorithm. Despite the fact that NEB is a well used method, it is all too often hard to use appropriately. Our investigations make it easier to make NEB's full capability available to newcomers approaching NEB for the first time. The following chapters use NEB and other popular techniques in the field to better explore and characterize promising new materials.
Another key focus of this work is the investigation of a coating composed of chromium and titanium (Cr-Ti), which, when applied to nickel maritime alloys, can mitigate mechanical and electrochemical corrosion. Using the NEB algorithm, we calculate relevant energy barriers for Ni, Ti and Cr surfaces to support future computational calculations that may be used to observe the evolution of the Cr-NiTi interface. Additionally, we computed the interfacial free energy of this interface and surface energies of individual Cr and NiTi systems, offering critical insights for macroscale modeling to predict continuum properties of the coating.
The final two chapters of this thesis address the acceleration of materials discovery by proposing new candidates for scientists, experimental or computational, to test. There are several online databases that make available the periodic structure of a wide variety of crystal structures. These databases can play a crucial role in generative tasks in producing new, previously undiscovered periodic materials. This generative process is realized by using a Generative Flow Network (GFlowNet) that sequentially samples the crystallographic properties of crystals, namely space group, composition, and lattice parameters, such that the physical constraints are obeyed. This approach also allows the sampling to be directly proportional to desirable properties of materials, as predicted by available property prediction models. This generation process is followed by an application of crystallographic symmetry and the concept behind Wyckoff positions to enumerate various atomic arrangements of crystals.This process is followed by computationally efficient structural optimization using a machine learning force field, resulting in potentially stable novel materials that are fully described in terms of their stoichiometry and atomic positions
DYSREGULATED IMMUNE RESPONSES TO VIRAL INFECTION AND VACCINATION
The understanding of immune responses to SARS-CoV-2 and other pathogens has been marked by significant research efforts aimed at investigating the complex mechanisms underlying disease and vaccination outcomes. Three studies in this thesis contribute to this body of knowledge by investigating the immunological aspects of Multisystem Inflammatory Syndrome in Children (MIS-C) following SARS-CoV-2 infection, the response of pediatric solid organ transplant recipients (pSOTRs) to COVID-19 mRNA vaccines, and the efficacy of a prophylactic hepatitis C virus (HCV) vaccine trial in people who inject drugs (PWIDs).
The first study investigates the pathophysiology of MIS-C, a syndrome affecting children following COVID-19. Utilizing advanced immunological assays, this research explores the role of emergency myelopoiesis, innate immune cell death, and specific cytokine profiles, notably IL-27, in distinguishing MIS-C from severe acute COVID-19. These findings not only offer insights into the mechanisms driving MIS-C but also propose potential diagnostic markers and therapeutic targets.
The next chapter focuses on immunogenicity of COVID-19 mRNA vaccines in pSOTRs compared to their healthy counterparts. Despite the immunosuppressive regimens, the study reveals that following additional doses of the COVID-19 mRNA vaccines, pSOTRs can mount humoral responses comparable to responses in healthy children. However, CD4+ T cell responses exhibit distinct profiles in pSOTRs, highlighting the need for tailored vaccination strategies to enhance protection in this vulnerable population.
The third study evaluates the efficacy of a prophylactic HCV vaccine in individuals at high risk of HCV infection due to injection drug use. Despite driving significantly lower HCV RNA levels in vaccinated individuals compared to placebo recipients, the vaccine did not prevent chronic HCV infection. The study identifies several baseline immune differences in PWIDs compared to people who do not inject drugs, including increased frequency of low-density neutrophils (LDN) and unique responses to the vaccine vector, suggesting that altered baseline immune states in PWIDs may influence vaccine efficacy.
Collectively, these studies enhance our understanding of dysregulated immune responses in the context of novel pathogens and vaccination, highlighting the nuanced interplay between the host immune system and immune-modulating forces. The thesis emphasizes the importance of considering individual immune status and environmental factors in developing effective treatment and prevention strategies against infectious diseases
12-Lead Electrocardiogram Anomaly Detection Using Hybrid 1D and 2D CNN+Transformer Architectures
This thesis presents two hybrid machine learning models for detecting anomalies in electrocardiograms (EKG): one that combines 1D convolutional neural networks (CNNs) with a transformer and another combining 2D CNNs with a transformer. The multilabel PTB-XL dataset includes over 21,000 cardiovascular conditions grouped into five superclasses: myocardial infarction, conduction disturbances, hypertrophy, ST-T wave changes, and normal EKGs. Due to its imbalance and multilabel nature, we implemented data augmentation techniques such as the Multi-Label Synthetic Minority Over-Sampling Technique (ML-SMOTE) and undersampling to handle the class imbalance. Additionally, the EKGs underwent advanced digital signal processing techniques to clean noise artifacts and convert the time-series signal into time-frequency space for two-dimensional representation. The 1D model achieved a 90.8% AUC, outperforming the 2D model, which achieved an 85.3% AUC