22689 research outputs found
Sort by
AI News Values: Axiologies Guiding Content Production in Google Gemini, Microsoft Copilot, and Newspapers
Full text linkThe rapid integration and cross-pollination of artificial intelligence (AI) into media production has fundamentally transformed how information is communicated, consumed, and understood. Despite this shift, limited research has explored the underlying news values encoded into algorithms used by AI chatbots that dictate news selection and determine how selected news stories are reframed through AI generated summarization. To explore differences in news content production in AI chatbots—Google Gemini and Microsoft Copilot—and newspapers–The New York Times, Los Angeles Times, and Iowa City Press-Citizen–a mixed methods approach utilizing content analysis, bibliometrics, thematic analysis, and computational grounded theory is conducted. The analysis illuminates two novel news values—efficiency and utility—exhibited in the news content AI chatbots generated. Findings suggest that news values exhibited by AI chatbots represent a departure from those in newspaper journalism–particularly lower circulation localized newspapers– and that underlying axiologies driving news content production in AI chatbots stem from a different set of objectives and constraints
Objective and Data-Driven Stillness Scores for PICU Patients
Full text linkOptimal sedation medication decisions are difficult to determine and personalize for individual patients and are resource-intensive. To address these issues, we propose accurately representing each patient's current state, allowing for the early detection of clinical anomalies and timely expert intervention. A key component is the sedation states of patients. Subjective nurse-conducted exams, such as the State Behavioral Scale (SBS), measure patient sedation at 2-4 hours low frequencies. We aim to create a higher frequency, fidelity, and data-driven sedation score. In our work, we have relabeled nurse-conducted SBS scores according to the small sample of ground truth scores. We will create a data-driven, comprehensive sedation score time series at a higher frequency by looking at multiple physiological data inputs, such as heart rate variability
Memory Through a Phase Transition: Smectic Focal Conic Domains in Microchannels
No full textIn this thesis, I have explored the effect of phase transition between different liquid crystalline phases on a particular family of topological defects called focal conic domains. In particular, I studied the behavior of focal conic domains in the smectic-A phase, how phase transition affects their optical properties and how these defects leave an imprint in the liquid crystal orientation, which can be retained even after a phase transition.
Smectic focal conic domains are geometric defects, which spontaneously occur in the smectic-A phase and which can be used as gradient-index microlenses. In this thesis, I studied the behavior of the defects when a chiral liquid crystal undergoes a phase transition from smectic to the cholesteric phase. After the transition, I observed that the defects could turn into a different type of defect, characteristic of the cholesteric phase. The cholesteric defects also act as micro-lenses but with a different focal length, therefore by confining liquid crystals in small vessels (microchannels), I created microlenses whose focal length is tunable with temperature. I identified the experimental conditions that make this possible, such as the concentration of chiral dopant and the rate of heating and cooling. I also explored the mechanism of transformation of these two types of defects into one another---a demonstration of memory at the phase transition.
Memory---the ability of LCs to ``remember'' information about the history of transformations they have undergone---has been demonstrated in many LC systems. I define geometric memory as the ability to reconstitute geometric defects (the focal conic domains) in the smectic-A phase after the system has been heated up through a phase transition and cooled back to the smectic-A phase. To understand geometric memory more thoroughly, I have designed an experimental protocol to quantitatively measure it and demonstrated that significant geometric memory is retained through the smectic-A--nematic phase transition and that the amount of memory depends on the maximum temperature at which the system is heated.
To make quantitative measurements, I developed an automatic image segmentation tool for the recognition of focal conic domains, previously used for galaxy detection.
In conclusion, I have developed new techniques to quantitatively describe geometric memory in liquid crystals and to understand under what conditions memory can be retained across a phase transition. I also utilized this memory effect to create tunable optical elements using liquid crystals
The Harmony of the Chaos. Teofilo Folengo's Chaos del Triperuno: a Book of Renaissance Crises
No full textTeofilo Folengo is the author of the mock epic poem Baldus, a lengthy epic crafted over the course of a lifetime. In 1527, following a radical shift in his life, which included leaving the monastery where he had spent ten years, Folengo moved to Venice and wrote the Chaos del Triperuno, a prosimetrum in three parts. In this work, he examines his life and processes the crises that shaped it. This dissertation undertakes a four-chapter examination of the languages, genres, and themes in Chaos del Triperuno to demonstrate how the poem’s structures, the crises it depicts, and its mixed-language style are not only symptomatic of Folengo’s personal religious and epistemological crisis but also reflect broader Renaissance crises. I argue that four central themes in the poem—soul, harmony, hell and doubt—interact with the religious and political turmoil of the first thirty years of the Cinquecento, highlighting how Folengo recorded the most problematic aspects of his time and tried to metabolize them. Despite its complexity and the enigmatic narrative scenes, it depicts, Chaos del Triperuno is a unicum in Folengo’s opera omnia and in the broader context of Italian Renaissance literature. The Chaos as a journey to truth, to Christ and grace, seems to be an educational process that leads one out of the maze of false beliefs and errors to a more authentic life. As Folengo would say: “Et ingannato al fine si ritrova chi lascia la via vecchia per la nova” (And he who leaves the old way for the new finds himself tricked in the end) (Chaos, R 381; Mullaney, 216)
Horace Made Strange: Carpe Diem as Mood, Reception as Attunement
No full textThis study examines Horace’s carpe diem odes, noting how landscape, atmosphere, and mood participate in poetic production. Roughly eighteen poems are usually grouped together as Horace’s carpe diem odes, though how this designation came about—and how it affects our understanding of Horace—has not been adequately studied. I offer a history of how this happened. I then suggest that these poems are aligned not only by convivial motifs, but also by a shared mood (Stimmung). However, far from reading carpe diem as a hedonistic imperative to enjoy the present, I argue that Horace’s poetry suggests that pleasure comes from an accumulation of the past. Turning to the aesthetic tradition of Stimmung, I further note how atmosphere and mood are enveloped in Horace’s poetic longing for that past. This longing, I suggest, is like the engine driving classical reception itself. Taking up carpe diem as a heuristic device for how reception works, this dissertation then charts Horace’s diffuse and sometimes unseen influence on a series of modern figures who use his carpe diem odes to think through gender, queerness, climate, addiction, and poetic production itself. Among the figures studied are poets such as Alcaeus, Lord Byron, Ernest Dowson, and Hope Mirrlees; jazz musicians such as Johnny Mercer and Bill Evans; and twentieth-century philologists
DYNAMIC FAILURE OF HETEROGENEOUS BRITTLE SOLIDS: DEFECT-INTERACTIONS, REPRESENTATIVE LENGTH-SCALES, AND UNCERTAINTY QUANTIFICATION
No full textStudying the failure under impact of brittle solids such as structural ceramics and rocks has several real-world applications e.g. modeling of planetary scale impacts or design of protective impediments under ballistic loading. Brittle materials are often
heterogeneous on some scale(s) with stochastic ‘defects’, introduced during processing or material genesis. Under relatively low-pressures, certain defects are known to play
a critical role in the cracking-based failure mechanism by acting as nucleation sites and/or altering the stress fields around existing cracks. Explicitly modeling the inter-
actions between a large population of defects/cracks and their role in material failure can be computationally expensive. In addition, the results of structural scale impact simulations often depend on the choice of the discretization volume and performing
mesh-sensitivity studies can be computationally prohibitive.
Modeling behavior under a range of pressures and strain-rates often requires a ‘multi-mechanism’ approach resulting in complex modeling frameworks which can be
difficult to implement in computational solvers. These models also need several input parameters which are uncertain due to material stochasticity. In addition, the large input parameter space can be hard to calibrate. Simpler phenomenological models
exist that are free of many of these problems, but their input parameter space is often not physical. A connection between the ‘fitting’ parameters of a simple model and
the ‘physical’ parameters of a complex model can, therefore, prove beneficial.
In this work we propose solutions to the aforementioned challenges in dynamic failure modeling of heterogeneous brittle solids. First, a numerical damage modeling
paradigm is proposed in which the effect of defects on crack-growth is incorporated to predict microstructure-informed strength. This model is computationally cost-
effective, adaptable to statistical microstructures and easy to calibrate. Further, a non-local constitutive formulation-based mesh-size determination procedure for compressive strength prediction is proposed. Use is made of microstructural information
gleaned from CT scans in developing the mesh-size estimates. Finally, uncertainty quantification is utilized to establish a connection between the material parameters of a physics-based yet complex model and another popular but simpler model. The
latter model is used to estimate the sensitivity of an impact performance metric to uncertain physical parameters
SELECTIVE MEMORY: THE ORBÁN REGIME’S MEMORY POLITICS AND RUSSIA’S WAR IN UKRAINE
No full textA social group’s collective identity is often profoundly influenced by its collective memory. What a social group interprets from its past can convey the meanings and morals that gives it a unique identity in the present. A constituent part of collective memory concerns the politics of memory, which involves the process by which political agents shape collective memory. Political agents seek to shape collective memory for a variety of reasons — to resolve insecurities about the past, bolster unique identities, and even to justify policy decisions. This thesis attempts to advance our understanding of how the Orbán government has used political memory to make virtue of its policies vis-à-vis Russia’s war of aggression in Ukraine.
This thesis begins by introducing the field of political memory and by providing a summary of Hungarian history. The third chapter of this thesis advances a theory to describe how contemporary Hungarian memory politics have changed since 2010 in response to the Hungarian government’s increasingly hostile relations with its Western partners. This chapter situates the mnemonic treatment of the war in Ukraine within the Hungarian government’s own contentions with its Western partners at a time when Mr. Orbán began his “Eastern Opening” foreign policy. The fourth chapter of this thesis proposes and tests a methodology to capture application of political memory through the use of Machine Learning sentiment analysis and the capturing of historical analogies and references in mass media to negatively portray Ukraine and Western support for Ukraine. The findings of chapters 3 and 4 show the Hungarian government makes virtue of its policies regarding the war in Ukraine by using adversarial and ideologically-charged historical analogies to frame Ukraine and Western support to Ukraine’s war effort, and that these historical analogies are consistent with a contemporary Hungarian political memory rooted in former glory, self-victimization, and resentment
PHARMACOLOGICAL PINK1 ACTIVATION MITIGATES PARKINSON’S PATHOLOGY VIA MITOCHONDRIAL BIOGENESIS
No full textMitochondrial dysfunction and α-synuclein accumulation are central features of Parkinson’s disease (PD) pathogenesis. The transcriptional repressor PARIS (ZNF746), a downstream target of PINK1, contributes to PD by suppressing mitochondrial biogenesis through inhibition of PGC-1α. While PINK1-mediated phosphorylation of PARIS at serine residues (S323 and S613) primes it for degradation, its therapeutic relevance remains unclear. Here, we investigate the efficacy of MTK458, a pharmacological activator of PINK1, in human and mouse models of PD and examine the role of PARIS phosphorylation in mediating its neuroprotective effects.
Using dopaminergic neurons derived from A53T α-synuclein iPSCs, we show that MTK458 reduces PARIS levels, enhances mitochondrial respiration, decreases phosphorylated α-synuclein (pSyn-129), and improves mitochondrial biogenesis. However, these effects were abolished in isogenic lines harboring PARIS S323A/S614A double knock-in (dKI) mutations, despite improved mitophagy, indicating that PARIS phosphorylation is necessary for MTK458 function and that impaired biogenesis, rather than mitophagy, underlies mitochondrial dysfunction in PD.
In vivo, MTK458 preserved dopaminergic neurons, reduced pSyn-129 accumulation, and restored dopamine levels in the striatum of wild-type mice following α-synuclein preformed fibril (PFF) injection. These neuroprotective effects were absent in PARIS S323A/S620A dKI mice, further confirming that PARIS phosphorylation is required for MTK458–mediated neuroprotection in vivo. Behavioral analyses revealed that MTK458 significantly improved motor performance in PFF-injected wild-type mice but failed to do so in PARIS dKI mice. Despite restored mitophagy in both genotypes, MTK458 could not reverse mitochondrial dysfunction or α-synuclein pathology in PARIS dKI models, suggesting that mitochondrial biogenesis—not mitophagy alone—is essential for therapeutic rescue. Together, our findings highlight the critical role of PINK1-mediated PARIS phosphorylation in regulating mitochondrial homeostasis and dopaminergic neuron survival. MTK458 represents a promising therapeutic approach for PD, but its efficacy is contingent upon the ability to degrade PARIS. These results underscore the importance of targeting both mitophagy and mitochondrial biogenesis to achieve effective neuroprotection in PD
Lifestyle medicine implementation: Teaching healthcare professionals to recommend healthy behaviors
No full textBackground. Lifestyle Medicine (LM) is an emerging medical specialty that can prevent or treat chronic diseases that are the leading causes of sickness and death in the US. Healthcare professionals report low knowledge and insufficient training as barriers to practicing LM. There is some evidence that continuing medical education may address these barriers, but additional research is needed to fully understand the implementation and impact of LM educational strategies.
This dissertation research addresses this gap by identifying, describing, and evaluating LM education strategies to facilitate LM implementation. Specifically, this dissertation work aimed to identify and describe current educational strategies for LM implementation in health systems and investigate how LM knowledge, confidence, and practice is impacted by LM education.
Methods. Data were collected as a part of two parent studies. The Lifestyle Medicine Integration in Health Systems: A Case Study Project was an exploratory, cross-sectional collective case study investigation. In-depth interview transcripts were analyzed identify and describe educational strategies to facilitate LM implementation.
The LME Program Evaluation Study included longitudinal analysis of survey responses from healthcare professionals participating in a LM educational program. Surveys evaluated self-reported LM knowledge, confidence, and practice behaviors and investigated changes before, after, and 8-12 months following program completion.
Results. The multiple case study analysis revealed that healthcare professionals especially need training in the topics of nutrition and behavior change counseling and that interpersonal learning is a critical educational strategy to LM adoption. Survey results indicated that completion of the LME Program is associated with increased LM knowledge, confidence, and practice and that changes are sustained 8-12 months following completion. They also suggested that individual determinants, such as professional training and LM certification status, were more associated with changes in LM knowledge and confidence, and organizational determinants, such as LM implementation climate and use of value-based healthcare practices, had stronger association with changes in LM practice.
Conclusions. Short, asynchronous, and online educational programs can be effective interventions for facilitating LM practice, especially effective when directed and novice healthcare professionals and accompanied by organizational support. Health system leaders should encourage broad participation in such educational programs
Comparison of Efficiency of Several Stochastic Approximation Algorithms
Full text linkThis work focuses on the comparison of efficiency of three stochastic ap-
proximation (SA) methods: simultaneous perturbation SA (SPSA), random
direction SA (RDSA), and truncated Cauchy smoothed functional algorithm
(TCSF). The derivation of asymptotic normality and mean square error (MSE)
of SPSA and RDSA is reviewed, and comparison shows that TCSF has asymp-
totically biased estimator, and hence the claim that it outperforms SPSA is
invalid. Modified ways of comparing the MSEs between SPSA and RDSA
is studied, and results show that for general loss functions SPSA tends to
outperform RDSA not necessarily deterministically but with a probability
exceeding 1/2