22689 research outputs found
Sort by
Using Polygenic Risk Scores in Risk Prediction: From Health Care Costs to Absolute Risk
Get PDFPolygenic risk scores (PRSs) are powerful tools that summarize cumulative additive effects of genetic variants, and have had increasingly more applications to risk stratification and clinical decision making with regards to chronic diseases. However, limitations to the use of polygenic risk scores include a lack of focus on a broader set of health outcomes, rather than disease incidence; the potential ramifications of PRS estimation uncertainty to affect clinical decision making; and the disparity in the efficacy of risk prediction in diverse populations, which could lead to exacerbating health inequities. Here, we analyze the joint association between PRS of a broad set of common diseases and related risk factors in the Atherosclerosis Risk in Communities study with health care expenditure and show that polygenic predisposition can predict future inpatient health care costs for males and females. We further explore the uncertainty in PRS construction by providing a framework for determining the impact of the posterior standard deviation of PRSs derived from Bayesian algorithms in clinical decision making, and find that such uncertainty makes little impact in the ranking and selection of individuals based on breast cancer risk within UK Biobank European ancestry women. Finally, we propose a flexible model for evaluation of the absolute risk of a disease and mortality taking into consideration polygenic predisposition, self-identified race and ethnicity, and genetic ancestry. Application of the model to All of Us individuals reveals importance of contextualization of risk using our framework to understand the true impact of the differential burden of polygenic risk across individuals of diverse background. Overall, this thesis advances understanding of the complex impact of polygenic predisposition to diseases and traits on the entire life-course of individuals, going beyond disease incidence and taking into account racial, ethnic and ancestral diversity
PRIMETIME POLARIZATION: BROADCAST MEDIA AND THE GROWING DIVIDE IN AMERICAN POLITICS
Get PDFWhile it is clear through voting patterns, evolving rhetoric, and increasing political extremism that American politics has become more divided in recent years, the question still remains as to why. As partisan politics split more to the right and left, abandoning the middle ground, many areas of research seek to highlight this growing issue; but there’s still room to add to the literature focused on the causes of such polarization. One area left untouched is broadcast media. This paper explores the relationship between broadcast media and political polarization, showcasing how various programs and events televised to American audiences can both highlight and influence the growing partisan divide across the country. Building off previous research, this paper examines the relationship between these two factors in depth, using data and examples from the past 12 years to provide a more modern and contemporary analysis. The three chapters in turn focus on recent election years and the different ways political information is shared via television - including news, televised presidential debates, and national nominating conventions. This work supports the argument that there is a connection between broadcast media and political polarization, with politicians and political pundits using this form of communication as a free platform to spread their message, rhetoric, and opinions - in turn, influencing the American public while also shining a spotlight on the growing divide. This is significant in that by determining possible causes of polarization, solutions can be found to prevent its further spread and radicalization
Applying TPACK-21 CQL Through Communities of Practice to Improve Teaching
Get PDFThis study examined the impact of a 10-week intervention using the TPACK-21CQL and 21st-century learning frameworks to enhance teachers’ knowledge and self-efficacy in integrating 21st-century skills within a Confucian heritage culture school. Using a convergent parallel mixed-methods design, the study analyzed quantitative and qualitative data collected before and after the intervention. Quantitative data included pre- and post-intervention surveys measuring constructs such as reflective learning (RL), authentic learning (AL), collaborative learning (CL), and design disposition (DD). Qualitative data from session transcripts and focus groups revealed themes related to the intervention’s effectiveness, challenges faced, and the impact on teachers’ professional development.
The findings indicated a significant change in constructs RL, AL, CL, and DD, suggesting the intervention enhanced teachers' TPACK-21CQL and ability to integrate technology and 21st-century skills. However, Design Thinking Efficacy (DTE) and Teachers as Designers (TaD) showed no statistically significant change, highlighting the need for ongoing professional development
NATO & THE DEMOCRACY DEFICIT: AN OLD PROBLEM WITH NEW RELEVANCE
Get PDFSince its inception in 1949, NATO has made a values-based approach central to its strategy, with its founding treaty linking alliance members to democracy, individual liberty, and the rule of law. Furthermore, since 1995 NATO has required that any nation seeking membership must fulfill certain criteria – “[including] a functioning democratic political system based on a market economy; fair treatment of minority populations; a commitment to resolve conflicts peacefully; an ability and willingness to make a military contribution to NATO operations; and a commitment to democratic civil-military relations and institutions” (NATO 2016). However, the treaty gives almost no guidance on how to best uphold these democratic precepts to states already within the alliance, and no defined process exists for ejecting or restraining a member state that fails to adhere to the values required of applicants.
In building a strategy that best positions NATO for success, this study asserts that a values-based approach will keep NATO member-states aligned with common interests and highlight the moral asymmetries between liberal democracies and autocratic regimes. However, recent democratic backsliding by key alliance members complicates this task. This study will assess 1) the extent to which democratic backsliding is a threat to NATO’s core mission and 2) potential options for dealing with it. To do so, it will draw primarily from three case studies – Portugal’s Estado Novo dictatorship (1949-1974), the Greek Junta (1967-1974), and Türkiye during various points of the Cold War (1960 coup d'état, political violence 1976-80 and military junta rule until 1983).
This study will also analyze the balance between hard and soft power that has enabled successful NATO strategy. For NATO, hard power is most closely linked to the alliance’s collective military strength and commitment to mutual defense outlined in Article 5; while soft power features in the Atlantic Treaty’s Article 2 – which obligates member states to strengthen their free democratic institutions. Ultimately, the study argues that these twin pillars of alliance strategy are mutually reinforcing, and that a failure of members to live up to NATO’s stated values can also pose a threat to its collective defense against an enduring military threat
CONTROLLING PLASMONIC AND ORGANIC SEMICONDUCTOR PHOTORESPONSES THROUGH INTERFACIAL MODIFICATIONS
No full textInteractions between light and direct bandgap semiconductors are responsible for energy harvesting in photovoltaic devices. Solar energy conversion efficiency in these devices is inherently limited due to extreme thermalization losses of incident photons higher than the bandgap of the semiconducting material. Incorporation of organic semiconductor materials capable of singlet fission into photovoltaics has the potential to double device photocurrent by generating two triplets for energy harvesting for every one incident photon. Practical device performance also depends on the ability of carriers to diffuse through the entire thickness of the semiconductor thin film, however thicker films are superior at absorbing photons at the peak of the visible solar spectrum. To overcome this challenge, plasmonic materials can be combined with semiconductors to increase visible light absorption and trapping inside optimally thin film solar cells. In this work, various methods including novel plasmonic aerosol optical detection and femtosecond transient spectroscopy are used to probe how interfacial interactions can be used to manipulate the optical properties and dynamics of plasmonic and organic semiconductor materials.
Chapter 2 presents experimental methods for the generation and steady-state optical detection of plasmonic aerosol particles. These methods are used in the work described in Chapter 3 to investigate how the optical properties of silica@Au nanoshells vary in the aerosol phase compared to colloidal nanoparticles. This study demonstrated effective generation of plasmonic aerosols with tuned plasmon resonances reflecting the change in surrounding solvation environment from to solution to the aerosol phase. Ligand-solvent interactions proved to play a key role in the degree of solvent retention and peak shifting in plasmonic aerosols, suggesting that particle functionalization and suspension solvent could be varied to modulate solvent shell thickness, therefore impacting the energy of plasmon resonances and access to the surface for airborne sensing and catalysis. In Chapters 4 and 5, the impacts of changing molecular packing on the ultrafast dynamics on a series of phase pure perovskite templated 6,13-bis(triisopropylsilylethynyl)pentacene (TIPs-pen) thin films are presented. Ultrafast spectroscopic studies reveal that templating significantly impacts singlet fission as illustrated by the template-dependence of the rate constants for singlet and triplet kinetics
Sculpting Topological and Optical Architectures in Nematic Liquid Crystals with Photoalignment
No full textTopological defects are universal patterns of nature. They affect a range of physical phenomena, from the formation of the early universe to the mechanical behavior of materials. Defects in nematic liquid crystals arise naturally, making them an ideal experimental platform for investigating one-dimensional linear defects, known as disclinations. Disclinations can be precisely induced in a nematic by manipulating the alignment of liquid crystal molecules at confining surfaces. At the vanguard of controlling the molecular order of liquid crystals is photoalignment, a technique that utilizes light to create precise, complex patterns in liquid crystal order. In addition to its role in fundamental research, photoalignment is finding practical applications in developing advanced optical devices, display technologies, and materials that respond dynamically to environmental stimuli.
This thesis employs photoalignment to investigate the physical behavior of disclinations in nematic liquid crystals. By integrating experimental methods with numerical simulations, we develop a set of design principles that enable quantitative predictions for the connectivity and shape of disclination lines. Our findings demonstrate how disclination architectures can be manipulated in situ, providing physical insights and practical tools for designing next-generation materials.
From an applied science perspective, this thesis also expands the capabilities and application scope of photoalignment in materials science and photonics. We demonstrate a technique to encode arbitrary three-dimensional orientational patterns of liquid crystal molecules on confining surfaces, validating the robustness of this approach by crafting refractive index gradient optics. This enhanced capability of photoalignment, coupled with our ability to create arbitrarily shaped disclination architecture, opens new opportunities for designing "smart" materials with potential applications in next-generation photonics, programmable soft materials, and responsive systems
Home Visiting Reach and Engagement of Pregnant Women Who Screen Positive for Substance Use Risk
Get PDFBackground: Evidence-based home visiting (EBHV) is a strategy for supporting expectant families and families with young children to promote healthy family functioning, positive parenting, and child health and development. EBHV has the potential to serve many families with substance use issues. However, reaching and engaging these families presents unique challenges. Understanding how EBHV programs reach and engage families with substance use issues is key to fulfilling its potential to support such families by meeting their needs and improving family outcomes. This multimethod dissertation investigates the extent to which EBHV services in New Jersey reach and engage pregnant women who screen positive for substance use risk.
Methods: Aims 1 and 2 quantitatively examined differences in reach and engagement indicators by substance use risk status using multilevel multivariate logistic regression models fit on data from the statewide Central Intake system. Aim 3 used reflexive thematic analysis to qualitatively explore EBHV engagement experiences among 11 women identified as positive for substance use risk by their home visitors.
Results: Aim 1 results indicate Central Intake was overall more likely to attempt to contact and refer women to EBHV who screened positive for substance use risk prenatally than those who screened negative. There were no overall significant differences in contact success or enrollment by substance use risk. However, interaction analyses highlight how housing stability, race/ethnicity, parenting experience, and social support modify the association of substance use risk with reach indicators. In Aim 2, there was not a statistically significant difference in receipt of a high dose of services by substance use risk status. Aim 3 results highlight the importance of trusted referral sources, tailored provision of functional supports both related and unrelated to substance use recovery, and trusting relationships with home visitors as key to participants’ engagement.
Discussion: This dissertation’s findings suggest that New Jersey EBHV’s efforts to reach and engage women with substance use issues were successful in some areas, while identifying areas for improvement. Grounded in the Home Visiting Precision Paradigm, discussion focuses on program design and implementation strategies to improve reach and engagement and ultimately to strengthen program effectiveness
ELECTRODYNAMICS OF TOPOLOGICAL INSULATOR THIN FILMS REVEALED BY TIME-DOMAIN THZ SPECTROSCOPY
Get PDFTopological insulators (TIs) are a special class of quantum materials that exhibit insulating behavior in their bulk while supporting conducting states on their surfaces or edges. These surface states arise due to strong spin-orbit coupling and are protected by time-reversal symmetry, leading to robust, spin-polarized, and massless Dirac fermions that are immune to non-magnetic impurities and disorder. The unique electronic properties of TIs, such as spin-momentum locking and topologically protected surface states, have garnered significant interest for potential applications in spintronics, quantum computing, and low-power electronic devices. A particularly intriguing phenomenon associated with TIs is the quantum anomalous Hall effect, where a quantized Hall conductance occurs without an external magnetic field due to intrinsic magnetic ordering induced by magnetic doping.
A high-sensitivity THz polarimetry setup was developed, achieving a polarization measurement precision of 0.02 milliradians. This was accomplished using a fiber-coupled laser-based THz spectrometer and high-extinction-ratio wire-grid polarizers, alongside a comprehensive calibration scheme to account for systematic errors.
This technique was applied to magnetically doped TIs to study the quantum anomalous Hall effect. While THz spectroscopy measurements of Hall conductance hysteresis loops qualitatively agreed with dc transport data, the frequency dependence contradicted known theoretical models for TIs with a clean magnetic gap. A spatially inhomogeneous magnetic gap model was used to interpret the optical response, attributing the discrepancies to doping inhomogeneity, likely caused by chromium atom clustering on the TI surface.
Further investigations into non-magnetic TIs revealed sub-quantized Hall conductances, attributed to low electron mobilities. Additionally, we observed large discrepancies in charge carrier densities derived from longitudinal and Hall conductance measurements. This was linked to spatial fluctuations in the Dirac point due to non-uniform doping, leading to electron and hole puddle formation in the surface states.
These findings underscore the significant impact of doping disorder and magnetic inhomogeneity on the electrodynamic properties of 3D TIs. Addressing these material challenges is crucial for realizing robust quantum Hall effects and advancing the application potential of TIs in quantum technologies
DISEASE MODELING AND MODIFICATION IN CELLULAR MODELS OF BARTH SYNDROME
Get PDFBarth syndrome (BTHS) is a rare, X-linked inborn error of mitochondrial phospholipid metabolism caused by pathogenic variants in the gene TAFAZZIN (TAZ), which leads to abnormal cardiolipin (CL) metabolism on the inner mitochondrial membrane. Although TAZ is ubiquitously expressed, BTHS involves a complex combination of tissue specific phenotypes including cardiomyopathy, neutropenia, skeletal myopathy, and growth delays, with a relatively minimal neurological burden. While the primary genetic and metabolic defects are well defined, there is limited mechanistic understanding of how pathogenic variants in TAZ, and therefore defects in CL metabolism, contribute to the mitochondrial pathogenicity of BTHS. Thus, there are limited targets for therapeutic monitoring and treatment.
To understand both the developmental and functional effects of TAZ-deficiency in different tissues, we generated isogenic TAZ knockout (TAZ-KO) and WT cardiomyocytes (CMs), skeletal muscle cell types (SKMs), and neural progenitor cells (NPCs) from CRISPR-edited induced pluripotent stem cells (iPSCs).
In TAZ-KO CMs we discovered evidence of dysregulated mitophagy including dysmorphic mitochondria and mitochondrial cristae, differential expression of key autophagy-associated genes, and an inability of TAZ-deficient CMs to properly initiate stress-induced mitophagy. Similarly, TAZ-deficient skeletal muscle cell types demonstrate failure to upregulate myogenic transcription factors and show preliminary evidence for dysregulated mitophagy. Further, in TAZ-deficient NPCs we identified novel phenotypes including a reduction in CIV abundance and CIV activity in the CIII2&CIV2 intermediate complex.
Using nutritional fatty acid supplementation strategies to manipulate CL acyl content, we discovered, while CL acyl chain manipulation was unable to alter mitophagy defects in TAZ-KO CMs, linoleic acid or oleic acid supplementation was able to partially restore CIV abundance in TAZ-deficient NPCs.
Taken together, our results have implications for understanding the tissue-specific pathology of BTHS and the potential for tissue-specific therapeutic targeting. Moreover, our results highlight an emerging role for mitophagy in the cardiac pathophysiology of BTHS, as well as in other skeletal muscle cell types, and we hypothesize that defective mitophagy could provide the missing link to unify the prenatal and postnatal cardiac complications in BTHS. Interestingly, we also reveal a potential neuron-specific bioenergetic phenotype and highlight the importance of evaluating downstream cellular defects in a tissue-targeted manner
Towards Efficient Long-Context Natural Language Processing
Get PDFTransformer-based language models (LMs) have achieved impressive performance on various natural language processing (NLP) benchmarks and applications. However, their efficiency and effectiveness in long contexts are still an open question. On the one hand, transformers suffer from their quadratic complexity in time and space. On the other hand, even with sufficient resources, LMs are unable to effectively comprehend long inputs or generate long and coherent texts. In this thesis, I tried to address both challenges by
1) investigating and diagnosing the existing methods on long-context LMs;
2) studying alternative solutions for language representation;
3) proposing a language encoding solution that is both efficient and effective.
Previous work mainly focuses on the efficiency of transformer-based LMs with numerous transformer variants. However, most of them cannot be directly adapted to NLP tasks. In Chapter 3, I propose methods to fill the gap between these LM prototypes and NLP applications, and my investigation surprisingly showed that most of them do not have advantages in long-context natural language understanding (NLU) tasks, suggesting that alleviating the computational overhead alone may not deliver satisfactory results.
In Chapters 4 and 5, I study a basic yet challenging problem: How can we find an alternative language representation method beyond the token-wise embedding paradigm? I start with SpanFinder, which encodes text at the span level and can be applied to NLP tasks such as semantic role labeling and FrameNet parsing. Taking a step further, I propose Soft Prompts, a method that learns non-text language representation to elicit factual knowledge from LMs. It automates the prompt engineering problem with gradient descent. My research shows that language representation does not necessarily correspond to individual tokens or specific texts.
With these findings, I propose to scale LMs to long context with Nugget in Chapters 6 and 7, which is a language representation solution that encodes text with variable lengths of vectors. Nugget fundamentally breaks the correspondence between tokens and vectors, which enables LMs to have a dynamic resolution for text representation and thus achieves efficiency in long texts. When applied on encoder-decoder transformers, Nugget can achieve nearly lossless encoding with much less memory overhead while achieving comparable or better performance on textual similarity tests. Dodo, the extension of Nugget on decoder-only LMs, can greatly reduce the time and space overhead of large language models (LLMs) such as LLaMA. Both models strongly imply that transformer-based LMs are so flexible that efficient text encoding with dynamic resolution is a viable way toward long-context NLU