22689 research outputs found
Sort by
The Cost of Inaction: How Permitted Harms and Moral Distress Impact Abortion Care Workers and Patients in Safe-Haven States
Get PDFFollowing the Supreme Court’s decision to overturn Roe v. Wade, numerous studies have examined levels of moral distress among healthcare professionals, largely focusing on those practicing in states with total or near-total abortion bans. Though vital, this research often overlooks abortion care providers in states where abortion protections remain. As a result, moral distress in these settings is underexplored and underdiscussed. Safe-haven states—states where access to abortion care has been enshrined in state legislature—are paramount for ensuring reproductive healthcare remains accessible in a post-Dobbs era. This thesis examines the conditions under which moral distress may occur for abortion care providers in safe-haven states, positing that they differ in nature from those present in restricted states. An ethical analysis approach was utilized to examine the current reproductive healthcare landscape post-Dobbs, as well as a real-world case that highlights the challenges faced by clinics providing procedural abortion services. The analysis revealed that abortion care providers in safe-haven states experience unique forms of moral distress, relating to resource constraints and witnessing patient harms. This thesis then suggests that safe-haven states may have additional moral and ethical obligations concerning abortion protections post-Dobbs. Ultimately, safe-haven states must develop a more nuanced understanding of abortion rights, shifting from solely protecting negative rights to the inclusion of positive rights, ensuring access to care is attainable and sustainable in a post-Dobbs world
Integrating Deep Learning Techniques in Radiology: From Classic Models to Foundation Models
No full textRadiology is a keystone in modern medicine, allowing for the accurate diagnosis, treatment planning, and monitoring of diseases using various state-of-the-art imaging modalities including MRI, CT, and X-ray. Despite the important role it plays, traditional radiological workflows are beset by inefficiencies, variability in accuracy, and the ever-increasing complexity of imaging data. These limitations point to an urgent need for innovative approaches to enhance workflow efficiency, standardization, and diagnostic precision.
This thesis is aimed at the transformative potential of AI to solve these challenges with three primary objectives. First, it works on deep learning models to extend radiological image analysis by developing a novel framework for cardiac coronary artery segmentation. The method contributes to the issue of disconnected points in segmentation, optimizing vascular topology and improving diagnostic accuracy in cardiovascular conditions. Secondly, it deals with the integration of radiological imaging with clinical data using foundation models of vision and language to make the diagnosis of diseases multimodal. Certain applications involve classifying ovarian cancer using the Segment Anything Model and LLMs for automatic CAD-RADS category assignment in coronary CT angiography. Third, the vision-language models are developed to bridge the gap between imaging data and textual analysis, advancing tasks such as automated radiology report generation, image-to-report analysis, and survival prediction. Key contributions include a 2D vision-language framework for cross-dimensional integration and a 3D VLM for CT pulmonary embolism radiology reporting.
These results from this research showcase how the latest AI technologies may disrupt radiology by making workflows more efficient, enhancing diagnostic accuracy, and personalizing patient care. By tackling basic challenges in medical imaging and radiological reporting, this thesis contributes to the development of precision medicine and opens the way for a new generation in radiological science
COMMERCIAL FOOD WASTE DIVERSION POLICY IN THE U.S.: ALIGNMENT BETWEEN SOCIOECONOMIC FACTORS AND POLICY DESIGN
Get PDFThe scale of commercial food waste in the U.S. poses a significant challenge with considerable climate and equity implications. Several jurisdictions have adopted policies to reduce commercial food waste and divert it from landfills. While much variation exists among current food waste diversion policies, they generally fall into one of three categories: recycling mandates, which require recycling of food waste; donation mandates, which require donation of edible surplus food and recycling of food scraps; and hybrid diversion policies, which require diversion of food waste but offer multiple pathways for compliance. This paper examines existing state-level commercial food waste diversion policies and explores the relationship between key socioeconomic factors and policy adoption. Specifically, this paper presents data on food insecurity, poverty, and income inequality in each jurisdiction with a statewide diversion policy in place, and it employs thematic analysis to determine if the degree of these socioeconomic factors correlates with the adoption of a particular policy type (e.g., do high levels of food insecurity correlate with adoption of donation mandates?). The findings of this research indicate that severe food insecurity does not consistently align with the adoption of any particular policy type, but high levels of poverty and income inequality appear to align with adoption of donation-based policies. This study offers recommendations for policymakers on integrating key socioeconomic factors into food waste diversion policy in a more consistent manner, and it identifies areas for future research to measure and enhance policy effectiveness
Development of Neuronal Connectivity in the Murine Neocortex
No full textIn this thesis, I explore a fundamental question in developmental neurobiology: how is ordered neuronal connectivity established in the brain? I use the murine somatosensory cortex as a model to first explore the coordinated emergence of neuronal subtype-diversity and spatial organization. Then I address the molecular and cellular mechanisms underlying the development of cortical neuron-subtype specific axonal projection patterns.
In the cerebral cortex, glutamatergic projection neurons are organized into layers based on their time of birth, as well as into areal domains in the plane perpendicular to the radial axis. In Chapter 2, I describe how the transcription factor Mef2c controls the acquisition of laminar and areal identities in post-mitotic cortical neurons during embryonic development.
Chapter 3 focuses on postnatal functions of Mef2c, identifying it as one of the first regulators of long-range intracortical axonal projection targeting. I then describe functional manipulations that demonstrate a role for EphA-EphrinA signaling downstream of Mef2c in mediating homotopic targeting of callosal projections. These observations offer a first glimpse into the molecular logic of interhemispheric projection targeting in the mammalian neocortex.
Chapters 4 and 5 focus on a critical aspect of specific intracortical connectivity: the formation of neuron-type specific patterns of intracortical axon collateral arbors. In Chapter 4, I introduce inducible sparse-labeling and genetic manipulation strategies developed in the Kolodkin laboratory to target developing neurons in cortical Layer 2/3, permitting the quantitative analysis of axonal collateral arbors at single neuron resolution. I then discuss their utility in uncovering novel cytoskeletal and cell-surface determinants of laminar specific Layer 2/3 cortical neuron axon branching.
In Chapter 5, I detail how sparse-labeling of Layer 6 corticothalamic neurons, combined with brain clearing and lightsheet microscopy, reveals the developmental dynamics of intracortical arbor elaboration by this understudied neuron subtype. This work sets the stage for future studies of molecular mechanisms that dictate divergent patterns of axonal elaboration, within the same target region, by distinct classes of cortical neurons.
My thesis work highlights the invariably pleiotropic nature of key regulators of animal development. This work also underscores the importance of temporally controlled, cell-type specific genetic access towards a comprehensive understanding of gene-function in development
RISK FACTORS FOR MORTALITY FOLLOWING AN ANAL CANCER DIAGNOSIS AMONG PEOPLE WITH HIV FROM 2010-2020 IN NORTH AMERICA
Get PDFAbstract
Background
Although with advanced treatment for anal cancer, people with HIV (PWH) still have a higher risk for cancer. While investigating the risk factors associated with anal cancer mortality among PWH was limited. This study illustrated demographic and clinical factors of all-cause mortality after anal cancer diagnosis in North America.
Methods
The study population was selected based on the cohort of the North American AIDS Cohort Collaboration on Research and Design (NA-ACCORD), which included 580 PWH diagnosed with anal cancer from 2010 to 2020. Using Poisson regression models and sensitivity analysis, we assessed the association between anal cancer mortality and both demographic and clinical factors together.
Findings
Among the study population, 29% died, followed by an anal cancer diagnosis. Regional/distant stage of cancer stage was strongly associated with increased risk of mortality in both unadjusted (RR=5.48, 95% CI: 2.28-17.98) and adjusted models (RR=3.77, 95% CI: 1.44-9.88). At the same time, every 10-year increase in age was only significant in the adjusted model (RR=1.42, 95%CI: 1.05-1.92). HIV viral suppression was also associated with increased mortality risk in the adjusted model (RR=1.42, 95% CI: 1.05-1.92). Nadir CD4 counts, Sex, race and ethnicity were not significantly associated with the outcome
DEVELOPING A MOUSE MODEL FOR EVALUATING SEX DIFFERENCES IN MYCOBACTERIUM AVIUM PATHOGENESIS
Get PDFThe incidence and prevalence of a group of nontuberculous mycobacteria (NTM) that includes Mycobacterium avium have been increasing in the US. These bacterial infections characteristically infect post-menopausal women who have had no history of lung disease, which led these infections to be described as “Lady Windermere Syndrome.” The cause of these clinical sex differences has not been well studied. We developed a mouse model using C3HeB/FeJ mice to evaluate sex differences in M. avium infection and pathogenesis. This mouse model is known for its increased susceptibility to mycobacterial infections. To investigate sex differences in the pathogenesis of M. avium, mice were infected with either MAV 101 and MAV 104, and monitored for morbidity, bacterial burden, and lung inflammation over 12 weeks. MAV 104 was effectively cleared from the lungs by 8 weeks pots-infection, while MAV 101 established a sustained infection. Surprisingly, despite sustained bacterial burdens with MAV 101, limited lung inflammation was observed in either sex. However, females exhibited significantly higher levels of IL-1α, a proinflammatory cytokine, compared to males, particularly after MAV 101 infection. Increased CCL2 levels were also observed at 4 weeks post-infection with MAV 104 and at 12 weeks with MAV 101. Furthermore, prior influenza infection exacerbated M. avium infection in males, but not females. This study demonstrates that MAV 101 is more pathogenic than MAV 104 in C3HeB/FeJ mice. Elevated IL-1α in females and increased CCL2 at 12 weeks suggest a host response to the bacterial infection, though whether these responses are protective is unclear. This work underscores the importance of considering strain-specific differences in M. avium pathogenesis and the potential impact of prior respiratory infections on NTM
susceptibility. Further research should investigate sex differences associated with M. avium pathogenesis in aged or gonadectomized mouse models to understand the role of menopause in the observed epidemiological data. Additionally, a primary insult model using a larger cohorts of mice should be considered. These studies should focus on pulmonary inflammation, cytokine responses, and histopathological analyses to better understand the mechanisms underlying sex-specific differences
COMMON GROUND EXPLORING NATURE, HUMANITY, AND PLACE ACROSS AN OCEAN
No full textNature is both an inspiration and a guide — wildly unpredictable yet beautifully ordered, fierce yet fragile. It is messy, hilarious, and awe-inspiring, an endless source of fascination. As a photographer, I’ve always been drawn to the details, compelled to zoom in and uncover the intricate mechanisms that make it all tick. That same curiosity now fuels my writing, shaping how I observe and interpret the world.
Living abroad has offered me a fresh perspective, shifting my lens both literally and figuratively. The contrast between familiar landscapes and foreign ones, between the ecosystems I once knew and those I now explore, deepens my understanding of how humans relate to the natural world. This thesis examines that relationship — how we shape nature and how it, in turn, shapes us — through the lens of landscapes an ocean apart
Value evaluation of a people-centered integrated care (PCIC) group in Shenzhen City, China
Get PDFBackground: In 2017, a people-centered integrated care group in Dapeng New District was established. It exemplified a cross-district model of "discipline alliance + compact medical community," representing a loosely structured, non-corporate governance consortium. This research aimed to evaluate this urban medical group through the triangular value chain.
Methods: The evaluation utilized the Donabedian model, focusing on three key dimensions: safety and quality, accessibility, and affordability. Longitudinal data was collected from 2017 to 2022 using government annual reports, the medical insurance bureau, and hospital information systems. Comparisons were made between pre-program and post-program outcome measurements to examine differences and trends. The effectiveness was notably assessable through the analysis of these trends.
Results: Accessibility assessments showed increased beds and medical staff per 1,000 people, including doctors, nurses, and pharmacists, along with a rise in referrals. Quality and safety assessments revealed a sharp increase in critically ill patients and complex procedures, while chronic disease management and patient satisfaction have improved. Affordability evaluations indicate growth in government subsidies and an acceleration in income from medical insurance. By 2021, medical insurance patients constituted 75.02%, up 44 percentage points from 31.19% in 2012. Bed utilization has also risen with more discharges, linking these outcomes to structural and process improvements.
Conclusions: Healthcare accessibility, safety & quality, and affordability in Dapeng New District have significantly improved. By transforming and delegating government functions, empowering primary healthcare is vital. Within the triangular value chain, prioritizing accessibility is essential, and integrating general practice with specialized services is key. The "Dapeng Model" can create exemplary outcomes and reduce disparities in basic health and health levels between urban and rural populations. Furthermore, we will pilot the payment reform and innovation within the Dapeng group to explore suitable payment models. The reform will accompany a health assessment and performance incentive system to encourage medical institutions to strengthen health management
Factors influencing the relationship between trust in the government and vaccine hesitancy among pregnant women in four countries (Brazil, Ghana, Kenya, and Pakistan)
No full textAbstract
Background: The Covid-19 pandemic has placed significant pressure on global health systems. The rapid rollout of the vaccine technology to curb viral transmission cast light on familiar issues of vaccine acceptance and the crucial role that factors like trust in the government and vaccine hesitancy play in public health. As a vulnerable subgroup, pregnant women face unique risks and uncertainties. Despite strong evidence supporting the safety and effectiveness of Covid-19 vaccines during pregnancy, vaccine hesitancy remained prevalent, posing potential risks to maternal and fetal health. Although trust in the government is increasingly recognized as a key determinant of vaccine acceptance, its influence on vaccine hesitancy among pregnant women remain underexplored in diverse global context. This study aims to measure the association between individual- level trust in government and covid-19 vaccine hesitancy among pregnant women in four countries Brazil, Ghana, Kenya, and Pakistan and to evaluate whether beliefs regarding vaccine safety and effectiveness modify this relationship.
Methods: A retrospective cohort analysis was conducted using cross-sectional survey data from pregnant women aged 18 and older who attended antenatal clinics and healthcare facilities between January 2023 and May 2024. The survey was used to capture detailed sociodemographic information, vaccination status and data for measuring key constructs of vaccine attitudes and beliefs, including vaccine hesitancy, trust in the government, and beliefs regarding the safety and effectiveness of Covid-19 vaccines. Multinomial logistic regression models were employed to measure the association between trust in government and vaccine hesitancy (primary outcome), adjusting for maternal age and educational and marital statuses. We separately evaluated the potential modifying role of perceptions of Covid-19 vaccine safety and effectiveness using interaction terms and post-hoc Wald tests.
Results: Low trust in government was associated with more than twice the odds of high COVID-19 vaccine hesitancy (aOR = 2.17, 95% CI: 1.62–2.91). Perceived vaccine effectiveness and safety significantly modified this association. Among participants with low trust and low/neutral beliefs about vaccine effectiveness, the odds of high hesitancy were significantly elevated (aOR = 2.04, 95% CI: 1.11–3.76). Similarly, individuals with both low trust in government and greater safety concerns had nearly six times the odds of hesitancy (aOR = 5.81, 95% CI: 2.35–14.40), while low trust alone became significantly associated with lower hesitancy in the absence of safety concerns (aOR = 0.36, 95% CI: 0.15–0.84). Beliefs about vaccine effectiveness alone remained a strong predictor of hesitancy even when not included in an interaction (aOR = 4.42, 95% CI: 3.32–5.88). Social influence and country of residence were also strongly associated with hesitancy outcomes across all models.
Conclusions: Findings underscore the complex, context-dependent relationship between trust in government and vaccine hesitancy among pregnant individuals. Perceptions of vaccine safety and effectiveness significantly modify this relationship, suggesting that trust in institutions alone does not uniformly predict hesitancy. Targeted public health strategies must therefore account for both interpersonal trust and individual-level vaccine beliefs when designing interventions to improve vaccine uptake in pregnancy. These findings support the need for localized, trust-building efforts and tailored communication that directly addresses safety and effectiveness concerns
Efficient Fault Injection for Exposing and Reproducing Failures in Cloud Systems
No full textIn today’s digital era, the reliability of cloud systems is paramount as the world increasingly depends on them for services. However, the complexity of modern cloud infrastructures, designed for high availability, fault tolerance, and scalability, makes the reliability of distributed systems a formidable challenge. Failures in these systems can lead to significant financial losses and other consequences, yet they remain difficult to detect, monitor, reproduce, diagnose, and recover from.
Faults, a common source of reliability issues in distributed systems, have attracted decades of research efforts. Many distributed system failures stem from bugs triggered by subtle faults under rare timing conditions. Furthermore, debugging failures often requires their reproduction, but reproducing fault-induced failures similarly depends on precise fault injection. This dissertation aims to improve the tools for fault-related bugs and failures, with a particular focus on bug detection and failure reproduction, to improve distributed system resilience. The core challenge is identified as how to efficiently determine the type, location, and timing of the injected fault. Firstly, a fault injection testing framework is proposed to efficiently expose partial failure bugs, by inferring and leveraging system state. Secondly, a tool is developed to efficiently reproduce fault-induced failures, by extracting and leveraging system runtime information. Lastly, this dissertation concludes by elaborating on the insights into future directions for further enhancing the efficiency of fault injection techniques to advance the state of reliability in modern cloud infrastructures