545298 research outputs found
Sort by
Time-Series of Inorganic Carbon Measurements on Surface Ocean Water near Bermuda and Hawaii, 1983-2023.
Get PDF
Understanding climate change anxiety and anticipatory climate disaster stress: A survey of residents in a high-risk California county during wildfire season
Get PDFBackgroundWith the increasing prevalence of climate-related disasters, psychological responses, including climate change anxiety and anticipatory climate disaster stress, have received heightened attention.ObjectiveWe investigate the correlates of climate change anxiety and anticipatory climate disaster stress, as well as the nature of these psychological responses.MethodsAt the start of the annual fire season (June to August 2023), we recruited a county-representative sample of n=813 residents of Lake County, in Northern California, to complete an anonymous online survey. Multiple regression analyses identified correlates of climate change anxiety and anticipatory climate disaster stress and explored how anxiety and stress were associated with disaster preparedness.FindingsClimate change anxiety, assessed via its cognitive-emotional impairment (odds ratio (OR)loss/injury=1.68; ORmedia=2.37) and functional impairment (ORloss/injury=1.68; ORmedia=2.63) subfactors, and anticipatory climate disaster stress (bloss/injury=0.15, bmedia=0.26) were associated with previous wildfire-induced loss/injury and media exposure to wildfire-related content. Anticipatory climate disaster stress was also associated with the frequency of being in an evacuation zone (b=0.05). Both the cognitive-emotional impairment subfactor of climate change anxiety (incidence rate ratio (IRR)=1.23) and anticipatory climate disaster stress (IRR=1.14) were associated with preparing an emergency kit and power outage supplies; anticipatory climate disaster stress was associated with evacuation intentions should an actual fire occur (b=0.12).ConclusionsPrior experiences with climate disasters could explain people's psychological responses to climate change. These responses could be temporally appropriate and functionally adaptive, given the immediacy of a potential fire.Clinical implicationsClimate change anxiety and anticipatory climate disaster stress should not be oversimplified as typical clinical symptoms because their presence might motivate adaptive self-protective behaviours in the face of an upcoming disaster
Development and Translation of Novel Hyperpolarized Carbon-13 MRI Acquisition and Post-Processing Techniques for Clinical Research
Get PDFResponse to therapy in advanced cancers is often unpredictable, and resistance to individual chemotherapeutic regimens or radiation therapy frequently occurs. The duration of time spent on effective systemic therapy is the most significant determinant of overall survival in patients diagnosed with nonresectable or metastatic disease at diagnosis. Therefore, there is a critical unmet need for accurate, quantitative metabolic biomarkers that can assess early therapeutic response for both research and clinical management. Hyperpolarized (HP) 13C MRI is a safe, noninvasive, and quantitative imaging technique that visualizes dynamic metabolic processes and can be used to detect and monitor treatment response of advanced cancers. In an approximately one-minute addition to an MRI exam, HP [1-13C]pyruvate MRI can detect cancer metabolic reprogramming through voxel-wise quantification of biomarkers, such as the first-order enzymatic conversion rates of pyruvate-to-lactate (kPL) and pyruvate-to-alanine (kPA). Changes in kPL reflect oncogenomic alterations associated with the progression or response of advanced prostate cancer to targeted therapies—such as PI3K, Androgen Receptor inhibitors (ARIs)—that CT, PET with FDG, C19-9 or PSMA cannot reliably provide. Thus, this dissertation focuses on the development of novel HP [1-13C]pyruvate MRI acquisition and post-processing techniques to overcome technical limitations and advance clinical translation. In the following chapters, I present projects from my graduate research that were designed to enhance HP-13C signal detection and to develop novel kinetic modeling techniques that incorporate additional biological characteristics to improve precision and accuracy for quantitative cancer studies. I also developed new MR acquisition strategies for increased resolution and clearer delineation of highly metabolically active lesions. Finally, I demonstrated the potential of deep learning (DL) methods to optimize k-space sampling reconstruction networks for highly accelerated HP-13C MR molecular imaging
Three-Dimensional Epigenomic Characterization of Human Brain Development
Get PDFGenome-wide association studies (GWAS) have identified thousands of non-coding variants that contribute to neuropsychiatric disease risks, likely by perturbing cis-regulatory elements (CREs). A significant barrier to understanding the genetic underpinnings of these neuropsychiatric complex diseases is the lack of functional characterization of risk genes and variants in biological systems relevant to human health. Moreover, as the human cortex is complex and heterogeneous, cell type-specific annotation of the 3D epigenome assists with insights into how non-coding genetic variants contribute to neuropsychiatric disorders. In the first chapter, I review how CRISPR interference (CRISPRi) and CRISPR activation (CRISPRa) screens can be leveraged to test non-coding variants associated with complex diseases. We first discuss the current challenges of interpreting the function of the non-coding genome and approaches to prioritizing disease-associated variants in the context of the 3D epigenome. Second, I provide a brief overview of high-throughput CRISPRi and CRISPRa screening strategies applicable for characterizing non-coding sequences in appropriate biological systems. Lastly, I discuss the promising prospects of using CRISPR-based technologies to dissect DNA sequences associated with neuropsychiatric diseases.
In the second chapter, we identified 3,489 and 3,894 functional CREs (fCREs) essential for iPSC fitness and cell survival during neuronal differentiation, respectively. These fCREs display dynamic epigenomic features and exhibit increased numbers and genomic spans of chromatin interactions following terminal neuronal differentiation. Furthermore, fCREs essential for neuronal differentiation show significantly greater enrichment of genetic heritability for neuropsychiatric diseases including schizophrenia (SCZ), autism spectrum disorders (ASD), and post-traumatic stress disorder (PTSD) than non-fCREs. Using high-throughput PRIME editing screens, we further identified 19 SCZ risk variants affecting cell survival during neuronal differentiation.
Lastly, in chapter three, I conducted a comprehensive 3D epigenomic analysis of four major glial populations, including ventricular radial glia (vRG), outer radial glia (oRG), oligodendrocyte precursor cells (OPC), and microglia (MG), from the mid-gestational human neocortex. By integrating gene expression, chromatin accessibility, DNA methylation, and 3D chromatin interactions, I identified cell type-specific candidate cCREs and validated their enhancer function using transgenic mouse embryos. Using machine learning, I prioritized 112 SCZ risk variants within glia cCREs and further confirmed the predicted vRG enhancer disruption by rs4449074 risk allele in vivo. Finally, oRG cCREs are enriched for human accelerated regions (HARs) compared to other cCREs. A subset of HARs have predicted activity differences compared to their chimpanzee orthologs and interact with genes involved in neuronal development.
In summary, this dissertation outlines the challenges and tools available to functionally evaluate disease associated variants with CRISPRi/a screens, provides a crucial resource for interpreting non-coding risk variants of neuropsychiatric diseases by extensive and in-depth functional annotation of cCREs during neuronal differentiation, and advances the understanding of human-specific gene regulation during corticogenesis
Competing signaling pathways controls electrotaxis
Get PDFAbstract:
Understanding how cells migrate following exogenous cues is one of the most fundamental questions for biology, medicine, and bioengineering. Growing evidence shows that electrotaxis, the directed cell migration toward electric potential gradients, represents a precise and programmable method to control cell migration. Most data suggest that the polarization of membrane components and the following downstream signaling are central to electrotaxis. Unfortunately, how these multiple mechanisms coordinate with the motile machinery of the cell to respond to an electric field is still poorly understood. Here, we develop a mechanistic model that explains and recapitulate electrotaxis across different cell types. Using the zebrafish proteome, we identify membrane proteins directly related to migration signaling pathways that polarize anodally and cathodally. Further, we show that simultaneous and asymmetric distribution of charged membrane receptors towards the anode and the cathode establishes multiple cooperative and competing stimuli for downstream signaling pathways. The resulting polarization of signals controls the actomyosin network dynamics, directing the anodal and cathodal migration of the cell. Our theoretical framework rationalizes the physical processes that determine electrotaxis across cell types and provides a physical framework to test multiple electrotactic pathways. These results together show us how to control cell migration to, e.g., enhance, cancel, or switch directed cell migration, which opens up new avenues not only to promote tissue regeneration or arrest tumor progression but also to design better biomimetic-engineered tissue constructs
Local Electric Field Effects on Water Dissociation in Bipolar Membranes Studied Using Core–Shell Catalysts
Get PDFThe local electric field strength is thought to affect the rate of water dissociation (WD) in bipolar membranes (BPMs) at the catalyst–nanoparticle surfaces. Here, we study core–shell nanoparticles, where the core is metallic, semiconducting, or insulating, to understand this effect. The nanoparticle cores were coated with a WD catalyst layer (TiO2 or HfO2) via atomic layer deposition (ALD), and the morphology was imaged with transmission electron microscopy. Irrespective of the core material, these core–shell catalysts displayed comparable WD overpotentials at optimal mass loading, despite the hypothesized differences in the electric field strength across the catalyst particle suggested by continuum electrostatic simulations. Substantial atomic interdiffusion between the core and shell was ruled out by X-ray absorption spectroscopy, X-ray photoelectron spectroscopy, and diffuse reflectance optical measurements. However, the optimal mass loading of catalyst was roughly 1 order of magnitude higher for the conductive and high dielectric core materials than for the low dielectric insulating cores. These findings are consistent with the hypothesis that electric field screening within the core material focuses the electric field drop between particles such that larger film thicknesses can be tolerated. Collectively, these data support the idea that it is the local electric field at the molecular level that controls proton-transfer rates and that the metal core/dielectric-shell constructs introduced here modulate that field. Further materials and synthetic design may enable optimization of the electric field strength across the proton-transfer trajectory at the material surface
Bootstrapping Large Language Model Robustness for Vision Language Model Safety via Reducing the Pretraining Modality Gap
Get PDFEnsuring Vision-Language Models (VLMs) generate safe outputs is crucial for their reliable deployment, but LVLMs suffer from drastic safety degradation compared to their LLM backbone, where blank or irrelevant images can trigger LVLMs to generate harmful responses to prompts that would otherwise be refused in text-only contexts. The modality gap between image and text representations has been recently hypothesized to contribute to safety degradation of LVLMs. This work shows that the amount of modality gap is highly inversely correlated with VLMs’ safety. It shows that this modality gap is introduced during pretraining LVLMs and persists through fine-tuning. Finally, this work proposes a regularization to reduce the modality gap during pretraining and validates it on LLaVA v1.5, ShareGPT4V, and MiniGPT-4, showing that this method substantially improves safety alignment of LVLMs
Empowering Language Models with Creativity
Get PDFCreativity is widely regarded as a hallmark of human intelligence and has historically propelled cultural and technological progress. The rapid development of artificial intelligence AI has prompted the question: Can machines exhibit comparable creativity? Recent advances in large language models (LLMs) offer promising tools to simulate creative activities through language. However, modern LLMs that primarily rely on in-domain fine-tuning still falter in low-resource, long-tail distributions, lack robust mechanisms for unconventional reasoning, and creative outputs remain challenging to evaluate due to their subjective nature and high-level attention required for verification.This dissertation presents a set of modeling, benchmarking, and evaluation techniques to bridge the gap in machine creativity expressed through language. The first part focuses on creative language generation, introducing techniques that disentangle training from inference to bypass the limitations of domain-specific data. Models are augmented at inference-time with domain knowledge and symbolic structures, enhancing their ability to generate humorous or artistically original outputs.The second part centers on unconventional reasoning, introducing a large-scale benchmark of cognitive puzzles that demand non-standard planning and tool use. These tasks probe LLMs’ capacity to reason beyond templates, compare human versus machine problem-solving, and propose techniques that allow models to refine their reasoning paths iteratively.The third part rethinks evaluation, advocating for idea-level analysis to move beyond surface metrics. By projecting outputs into latent event spaces or analyzing narrative arcs and trajectories at the discourse-level, this part proposes idea-level methods that better differentiate human and machine creativity.Together, these contributions build a foundation for creative language modeling—models that can generate, reason, and reflect more like humans. By addressing both generation and evaluation, this thesis aims to advance the field toward more imaginative, controllable, and user-engaging AI systems
Essays on Macroeconomics
No full textThis dissertation contains three essays on macroeconomics.The first two chapters examine how sudden stop episodes affect aggregate productivity through the reallocation of resources. During these episodes, domestic aggregate demand contracts while foreign aggregate demand remains largely stable, and exchange rate depreciation favors exporters. This shift leads to a relative expansion of export-oriented activities over domestic-oriented ones. Due to differences in market power and tax treatment, export-oriented activities exhibit lower revenue-based total factor productivity (TFPR) than their domestic counterparts. As a result, the reallocation of resources toward export-oriented activities reduces aggregate TFP. In the first chapter, I use detailed microdata from Mexico to document differences in distortions and resource reallocation at the plant–product–destination level during the 1994 sudden stop. In the second chapter, I build a simple model to clarify the mechanism and then develop a quantitative multisector small open economy New Keynesian model with input–output linkages. I show that reallocation effects alone account for approximately 50 percent of the observed decline in value added in the manufacturing sector.In the third chapter, coauthored with Saki Bigio, we reformulate the New Keynesian model to incorporate output adjustments through job flows—the extensive margin. Critically, we distinguish the stock and flows ofworkers available for production. Shifting from adjustments through the intensive to the
extensive employment margin, the model introduces predetermined output, altering key
properties of the New Keynesian framework. First, the Taylor principle is inverted: stability
is achieved when nominal rates respond less than one-for-one with inflation. Second, the
model significantly alters the output responses to changes in monetary policy. We argue
that this represents a challenge and an opportunity for the literature. Sticky information
allows the model to correct the sign of impulse responses