22689 research outputs found
Sort by
JOHNS HOPKINS HEALTH SYSTEM EMERGENCY PREPAREDNESS PERFORMANCE ASSESSMENT
Get PDFThis doctoral dissertation uses the workplace challenge format and was conducted in the 5 of the 6 hospitals of the Johns Hopkins Health System, varying across the sections of the study. The 6 hospitals are The Johns Hopkins Hospital, The Johns Hopkins Bayview Medical Center, Inc., Howard County General Hospital, Sibley Memorial Hospital, Suburban Hospital, and The Johns Hopkins All Children’s Health System in the United States. The dissertation’s primary aim was to identify the top hazards that each organization was at risk for and to assess their level of emergency preparedness to face these hazards. This information will help in structuring and prioritizing the emergency management program efforts to develop a comprehensive all-hazards approach for the Health System to prepare and respond to emergencies. It will also strengthen the emergency preparedness and response capabilities, thereby enhancing the resilience of Johns Hopkins Health System in the face of emergencies.
An organizational assessment of the Johns Hopkins Medicine Office of Emergency Management (JHMOEM) was performed in the areas of leadership, operations, workforce, measurement analysis, and knowledge management using the Baldrige Excellence Framework (Baldrige Performance Excellence Program, 2021). Surveys were used to collect the information from the key stakeholders of the JHMOEM.
The hazard vulnerability analysis results of the year 2021 of the five hospitals were used to assess the level of emergency preparedness of the Johns Hopkins Health System's hospitals for the top five hazards. The top five hazards each organization was at risk for and how prepared it was to face them were identified. Also, the top hazards that the Johns Hopkins Health System's Hospitals as a whole were at risk for were identified and prioritized.
The emergency operation plans and the hospital’s online policies of the five hospitals were reviewed in order to evaluate the hospitals' preparedness for the top five hazards specific to each hospital in relation to their preparedness scores. The lessons learned, role of the leadership, and the potential interventions in creating an all-hazard approach to emergency preparedness and response were mentioned in the discussion section
The Abolishment of the Arms Control and Disarmament Agency: Exploring the Change Agent and Primary Catalyst
No full textClaims that the Arms Control and Disarmament Agency’s (ACDA’s) abolition in 1997 was a mistake and that its functions should be reinstituted lead to the following questions: 1) Was the abolishment of ACDA in the 1990s unavoidable? 2) What were the key factors and who were the key contributors to the disestablishment of the Agency? and 3) Why was the agency with the independent mandate to address non-proliferation issues abolished at a time when nonproliferation was a high-priority goal of United States foreign policy? Utilizing the Bureaucratic Process Model (BPM) to study why the Agency responsible for negotiating and maintaining the U.S. arms control regime that had developed by the mid-1990s, through which non-proliferation efforts were coordinated, was abolished, adds to the current discussion regarding whether the reasons cited for its abolishment have continued relevance.
ACDA’s abolition was not the result of a series of seemingly random events. Rather, the events reflect the culmination of a pendulum “shift” in theory on the utilization of arms control as an agent to ensure U.S. national security, resulting in the elimination of the independent “voice” for arms control. While ACDA faced critics and challenges to its existence prior to the 1990s, prior efforts had been thwarted by ACDA’s supporters, within the executive and legislative branches. Most of the opposition to ACDA was led by Senator Jesse Helms, who, as an agent with both intent and influence outside the executive cabinet’s BPM, was able to outlast and outmaneuver Administration and Congressional opposition. Once Helms found partners within the BPM, in Madeline Albright, and within the opposition party on the Senate Foreign Relations Committee, in Joe Biden, the fate of ACDA was sealed. Helms was the change agent: His removal from the equation is the only formula modification that might have resulted in ACDA’s continued existence. The Chemical Weapons Convention (CWC) was the catalyst utilized by Helms to extract the concession of ACDA’s disestablishment from the Clinton Administration, which was desperate to ratify the CWC. In the absence of the CWC, Helms would have utilized another issue as his catalyst
Modeling Meaning for Description and Interaction
Get PDFLanguage is a powerful tool for communication and coordination, allowing us to share thoughts, ideas, and instructions with others. Accordingly, enabling people to communicate linguistically with digital agents has been among the longest-standing goals in artificial intelligence (AI). However, unlike humans, machines do not naturally acquire the ability to extract meaning from language.
One natural solution to this problem is to represent meaning in a structured format and then develop models for processing language into such structures. Unlike natural language, these structured representations can be directly processed and interpreted by existing algorithms. Indeed, much of the digital infrastructure we have built is mediated by structured representations (e.g. programs and APIs). Furthermore, unlike the internal representations of current neural models, structured representations are built to be used and interpreted by people. I focus on methods for parsing language into these dually-interpretable representations of meaning. I introduce models that learn to predict structure from language and apply them to a variety of tasks, ranging from linguistic description to interaction with robots and digital assistants.
I address three thematic challenges in modeling meaning: abstraction, sensitivity, and ambiguity. In order to be useful, meaning representations must abstract away from the linguistic input. Abstractions differ for each representation used, and must be learned by the model. The process of abstraction entails a kind of invariance: different linguistic inputs mapping to the same meaning. At the same time, meaning is sensitive to slight changes in the linguistic input; here, similar inputs might map to very different meanings. Finally, language is often ambiguous, and many utterances have multiple meanings.
In cases of ambiguity, models of meaning must learn that the same input can map to different meanings
Can Early SEZs be Drivers of Economic Growth in Transition Economies? Comparative Analysis: China, Russia, Vietnam
Get PDFThis research began with questions arising from disparate economic performance of early special economic zones (SEZs) in transitional economies at the initial phase of reform and market opening, specifically in China, Russia, and Vietnam despite their shared legacy of command economy as socialist states. Having a focus on the political economic dynamics of SEZ development, this comparative analysis examines how institutional governance structure of the respective countries impacted economic growth. It employs six key elements for SEZ development in the context of transition economies, chosen for their heuristic value: 1) state capacity for effective governance, 2) legal framework, 3) political stability, 4) infrastructure, 5) location, and 6) global networks, all of which are grouped into categories of internal foundation and external connectivity. Findings indicate that SEZs can serve as drivers of economic growth in the early stages of a socialist country’s transition to a market-oriented-economy when a host country with political stability and strong institutional capacity grants adequate autonomy and incentives to local authorities for effective governance of SEZs in strategic locations. Adequate infrastructure along with functioning state agencies that enable and promote coherent economic activities by both domestic and international actors with trust in the political commitments and stability of the host country were quintessential for SEZ development. Failure or success of a zone was linked to the efficacy of an institutional and incentive framework, whether it was strategically located both domestically and internationally, and the level of coordination between the central and local governments
COMPUTATIONAL ANALYSIS OF SINGLE-CELL TRANSCRIPTOMIC DATA FOR THE IDENTIFICATION AND CHARACTERIZATION OF CELL IDENTITY AND FATE POTENTIAL
Get PDFSingle-cell transcriptomics and the tools developed for single-cell RNA sequencing (scRNA-seq) analysis have enabled the characterization of tissues and disease states of interest, identification of rare cell types, and reconstruction of developmental lineages given only a snapshot in time. The field of transcriptomics has transformed from the first single cell analysis in 1990 to the development of high-throughput sequencing in the early 2000’s. Advances in computational analysis have paralleled development in the technology, though many analysis tools have ample shortcomings. Some of the challenges facing scRNA-seq data analysis will be addressed in this thesis, such as the development of a reliable method to quantify single-cell fate potential, the standardization of cell type annotation and downstream analysis, and the characterization of rare cell populations.
The objective of this work is to develop a computational pipeline for single cell transcriptomic data analysis which incorporates a novel tool to quantify single-cell fate potential, termed “stemFinder,” and to apply this pipeline to interrogate rare and clinically relevant cell types and to characterize key transformations in cell type identity throughout the course of physiologic differentiation or pathological transformation. In this work, I demonstrate the robustness of stemFinder to changes in input parameters and its superior performance to current methods of in silico potency quantification. I also use stemFinder to illustrate interesting biological concepts regarding cell cycle gene expression patterns and cell fate commitment and the relative potencies of distinct populations. stemFinder is readily incorporated into a scRNA-seq data analysis pipeline to elucidate the physiology of fallopian tube epithelial cell self-renewal and bone remodeling, as well as the pathophysiology of high-grade ovarian cancer and rheumatoid arthritis.
This thesis details the development and application of a scRNA-seq computational analysis pipeline—and create a novel tool as part of this pipeline—to create a census of the benign human ampulla and to implicate a rare cell type with a role in fallopian tube epithelial regeneration. This thesis will also leverage scRNA-seq analysis to understand osteoclastogenesis and investigate two subpopulations of anabolic and catabolic pre-osteoclasts, respectively, with implications in bone erosion and rheumatoid arthritis
ACTIVE LEARNING FOR LARGE-SCALE BOUNDARY LAYER WIND TUNNEL EXPERIMENTS
Get PDFActive Learning is a branch of Machine Learning (ML) that aims to optimize experimental design and reduce cost by minimizing the number of required experiments to achieve a given objective. In traditional ML, large datasets are required to train models effectively. However, acquiring data can be expensive and time-consuming. Active Learning algorithms address this challenge by iteratively selecting new samples that are predicted to be most informative. This kind of adaptive approach can focus experiments on estimating certain quantities of interest and converge to accurate results using a small number of experiments. This makes active learning very attractive for large-scale experimental setups such as those conducted in the Boundary Layer Wind Tunnel (BLWT). BLWT experiments are time-consuming, thus it is necessary to design each experiment carefully and select experimental configurations that are expected to be information rich. In this project, we apply active learning concepts to large-scale BLWT experiments by controlling the “Terraformer” – an automated surface roughness grid – at the University of Florida BLWT to design experiments aimed at learning the influence of terrain roughness on statistical properties of resulting near-surface atmospheric turbulence and the resulting pressures exerted on physical infrastructure.
We employ a novel active learning framework to identify relationships between stochastic roughness element configurations and second-order statistical properties of the generated wind velocity fields. These results prompt further exploration into understanding the impact of higher-order statistical properties on wind pressures. The learning framework is enabled by two automated tools unique to the UF-BLWT: the Terraformer roughness grid and a mechanized instrument traverse capable of collecting velocity measurements throughout the tunnel. Together these elements enable high-throughput experiments and rapid data collection for on-the-fly processing and active learning. Active learning is conducted with the use of Gaussian Process regression models and customized learning functions designed to target second-order statistical properties of the wind field. We specifically studied and customized three learning functions for this purpose – a modified version of the existing U-learning function previously used for assessing reliability, a modified version of the Expected Improvement for Global Fit (EIGF) learning function aimed at producing the best global regression model, and a new learning function termed the MUSIC (minimizing uncertainty in sensitivity index convergence) function aimed at rapid global sensitivity analysis (GSA).
Using the proposed active learning framework, 861 unique experiments are conducted with machine-learned roughness grids, over a period of 340 hours in 3 distinct phases. These experiments focus on simulating the terrain roughness using various 1-D longitudinal variations, ranging from simple sine wave shapes (phase 1) to triangular and square wave shapes (phase 2) and complex stochastic fields (phase 3). Each study identifies parameter ranges for the terrain roughness that yields second-order statistically equivalent wind profiles, along with the uncertainty in the prediction of this domain. These results are used to explore the influence of higher-order statistical variations within second-order equivalent wind fields on extreme wind pressures on critical infrastructure. This research project highlights the potential for active machine learning to improve, and potentially optimize experimental design and gather valuable insights from large-scale wind tunnel experiments that could not be achieved using traditional experimental design methods
THE SELF-PERCEIVED EFFICACY OF EDUCATORS AND THE IMPACTS ON READING PROFICIENCY FOR MULTILINGUAL LEARNERS
Get PDFMultilingual learners (MLs) are a growing population across schools in the United States. The increase in enrollment has presented a new challenge for educators, as they are tasked with supporting both content and language for culturally and linguistically diverse learners. Resources and language support for multilingual students can vary significantly, and some teachers are not prepared to meet their instructional needs within the general education classroom. Multilingual students may experience exposure to academic and socio-emotional stressors which can further impact the learning experience. The needs assessment revealed that middle and high school teachers were feeling low self-efficacy in their abilities to teach MLs within the classroom. When comparing the reading performance of multilinguals to their monolingual peers, the multilingual students demonstrated low reading proficiency on grade-level reading assessments. Additionally, the needs assessment determined that teachers would benefit from professional development on the topic of multilingual instructional scaffolding and support. Despite educators’ desire to serve their students, the low levels of self-efficacy presented a potential roadblock for ML student reading achievement. A five-session intervention study was designed to build educator capacity to implement brain-targeted instructional scaffolds to support multilingual students within the classroom. The intervention was integrated into the weekly Professional Learning Community (PLC) and participants had the opportunity to learn an instructional strategy each week and participate in professional discourse, reflection, collaboration, and guided planning. A convergent parallel mixed methods design was implemented to evaluate the process and outcomes of the designed intervention. Guided by Guskey’s (2002) Model of Teacher Change, the intervention resulted in a positive effect on teacher self-perceived efficacy as well as observed changes of instructional practice. Student reading proficiency scores demonstrated a statistically significant change after the intervention for both ML and non-ML learners
Molecular simulation towards energy applications: from traditional model to machine learning
Get PDFEntering the 21st century, it is imperative to deploy alternative energy solutions to increase energy availability and tackle climate change. Semiconductor materials rise as the backbone of renewable energy solution and they also provide a variety of applications based on their unique electronic, mechanical and thermal properties. By understanding the relevant growth and self-assembly process, we can tune the materials function on a molecular level and further improve the synthesis process for easy adaptation in commercial settings. Using a computational approach can vastly expedite this effect. This thesis demonstrates how we can use Molecular Dynamics and Density Functional Theory to investigate the thermodynamic and kinetic behaviors during the organic/inorganic semiconductor growth and how we bridge the atomic-level insights with experimental discovery. We also utilizes cutting-edge machine learning model to accelerate the materials discovery. We offered valuable guidance on how to develop an reactive and transferable force field based on on-the-fly Gaussian Process model. This force field allows us to study complicated crystallization process during the semiconductor additive manufacturing process which enables new route to fast and cost-efficient production. Lastly, this thesis steps further to data-driven approach to screen possible solar cell candidate using a graph neural network. Our work combines both traditional computational materials science approaches and state-of-the-art machine learning methods to bring us forward to a more versatile and sustainable energy future
The Dollar-Yen Exchange Rate: Appreciation Impact on Japan's Economic Cycles and Long-run Equilibrium between Its Deviation from Purchasing Power Parity and the Economic Performance
Get PDFThrough Vector Autoregression analyses, this paper examines the impact of USDJPY real exchange rate appreciation on the Japanese economic cycles since the floating regime started in 1973. Key testing variables are cyclical components of macroeconomic variables of output, trade, and asset prices. Long-run equilibriums between the deviation of USDJPY from Purchasing Power Parity (PPP)—i.e., the real exchange rate assuming that PPP holds—and the macroeconomic variables, if they exist, have economic implications. Thus, the paper also investigates their long-run relationships through cointegration tests. The test period is from 1991Q2 to 2007Q2 between Japan’s asset bubble collapse and the Global Financial Crisis, minimizing potential distortion by structural breaks.
This study finds that the USDJPY exchange rate does not function perfectly as a shock absorber. An appreciation shock has negatively and persistently impacted real GDP and consumption. At a minimum, the one-time shock of appreciation by 4 percent reduces real GDP by 0.2 percentage points. Given the yen’s historical swing, the adverse impact of the yen's appreciation could be significant relative to Japan’s sluggish growth (average annual growth after 1991Q2 is 0.7 percent). Unit labor costs are sticky in the short term and negatively cointegrated with the deviations from PPP in the long term. Given Japan’s low productivity over the decades, the result implies the correlation between the yen’s overvaluation and a wage cut. Exports and imports are irresponsive to an appreciation shock. Share prices demonstrate a persistent and positive response, implying the price rigidity of listed companies. Housing prices also show price stickiness. In the long term, the overvaluation puzzlingly correlates with a rise in housing prices despite the connection between overvaluation and a wage decline
Devil Dogs and the Valley of Death: Partisan Interests and Marine Corps Innovation
Get PDFWhat factors speed military innovations across the DoD's "valley of death?" Does service culture play a significant role in this process? Military innovation studies research continues to clarify complex factors like bureaucratic pressure, organizational culture, service rivalry, and technological change that drive the pursuit of warfighting advantage. In this study, I will employ a comparative, exploratory case study method relying on small-N qualitative research and process trace three purposive cases of Marine Corps service innovation using interpreted veto player and adoption-capacity theory. While these theories have been applied to state systems to produce causative covering laws of governance and bureaucratic change, they can be effectively interpreted to inform middle-range theory in military innovation. I propose that while institutional veto players ultimately control financial resources, organizational capital remains controlled by partisan veto networks that moderate military innovation. Echoing Tsebelis’ veto player theory in the case of policy stability in democratic systems, I theorize that partisan communities of interest can create a stymying effect that prevents disruptions to status quo warfighting practices that characterize true innovation. This paper further explores the organizational tension between communities of interest internal to the Marine Corps that moderate adoption rates for critical new programs and technologies