UTSA Runner Research Press (Univ. of Texas at San Antonio)
Not a member yet
6846 research outputs found
Sort by
Optimizing Efficiency of Optimistic Parallel Discrete Event Simulation with Deep Q-Networks
No full textOptimistic parallel discrete event simulation (PDES) leverages parallel computing to execute simulations, addressing the limitations of sequential simulation by improving execution time and memory efficiency, making it a viable technique for large-scale complex models. However, the dynamic nature of such simulations necessitates continuous optimization of critical parameters to minimize rollbacks and enhance performance. We evaluate the potential of predicting rollback behavior and other related dynamics of optimistic simulators through Deep Q-Learning, a type of reinforcement learning algorithm, to enhance overall efficiency. We experiment with a parallel discrete event simulation designed for logic circuits, which employs a shared parallel event queue where messages are inserted into local processor channels and managed using a rollback concept to handle discrepancies and optimize processing. Our research presents a novel method for autonomously optimizing critical PDES parameters, namely the number of threads, the time window size, and the number of buckets in the calendar queue, using a Deep Q-Network (DQN) framework. Our DQN model, a type of deep neural network, learns an optimal scheduling policy through reinforcement learning by interacting with the PDES environment. We define the state space with comprehensive simulation metrics such as the number of rollbacks and execution time. The DQN agent dynamically adjusts the simulation parameters based on these observations to achieve a balance between computational efficiency and simulation accuracy. Our results demonstrate that the DQN-based approach can significantly reduce rollbacks and improve execution time, offering a scalable and automated solution for optimizing PDES. This work not only advances the field of discrete event simulations but also provides a robust framework for future research in optimizing large-scale, complex simulations. Additionally, we explore a black-box approach named Bayesian optimization to predict the best parameter configuration, and its experimental findings are encouraging, particularly when we include rollback and execution time in the reward.Computer Scienc
Contribution of Astrocytes in the Inflammatory Response During Hypoxic Retinopathy
No full textThe full text of this item is not available at this time because the author has placed this item under an embargo until March 11, 2025.Diabetic retinopathy (DR) is the leading cause of vision loss in the working age population and a significant global health concern. Evidence suggests that chronic inflammation and hypoxia contribute to retinal degeneration. Despite advances in understanding how inflammation and microglia contribute to DR pathology the responses of astrocytes, glial cells essential for vascular development and neuronal homeostasis, are not well understood. Inflammatory signals induce a neurotoxic response in astrocytes while hypoxic signals induce anti-inflammatory and neuroprotective responses. Here we seek to understanding the contribution of hypoxic astrocytes to retinal pathology. To investigate the contribution of hypoxic astrocytes in retinopathy a model of systemic hypoxia was used with animal models of microglia dysregulation which exacerbates inflammatory responses. This allows us to investigate how hypoxia and inflammation affect astrocyte responses in vivo. Here we report that hypoxia induces an anti-inflammatory response with an increase in CCL17 and decrease in IL-36γ; and that hypoxia dampens inflammatory response to endotoxemia with reduced expression of IL-2 and TNF-α dependent on microglial regulation. We also report hypoxia does not exacerbate inflammatory responses in the GFAP-CreERT2:Rosa26iDTR mouse, a transgenic mouse that allows for DTx-induced inflammatory astrocyte activation. These results suggest that hypoxic signaling in DR induces an anti-inflammatory effect. As inflammation and hypoxia are a feature of CNS diseases and injuries such as stroke, traumatic brain injury, multiple sclerosis, Alzheimer’s disease, retinitis pigmentosa, and age-related macular degeneration the results presented here may be translatable to other disease research.Molecular Microbiology and Immunolog
Seasonal and Interannual Variations in Sea Ice Thickness in the Weddell Sea, Antarctica (2019–2022) Using ICESat-2
No full textThe sea ice extent in the Weddell Sea exhibited a positive trend from the start of satellite observations in 1978 until 2016 but has shown a decreasing trend since then. This study analyzes seasonal and interannual variations in sea ice thickness using ICESat-2 laser altimetry data over the Weddell Sea from 2019 to 2022. Sea ice thickness was calculated from ICESat-2's ATL10 freeboard product using the Improved Buoyancy Equation. Seasonal variability in ice thickness, characterized by an increase from February to September, is more pronounced in the eastern Weddell sector, while interannual variability is more evident in the western Weddell sector. The results were compared with field data obtained between 2019 and 2022, showing a general agreement in ice thickness distributions around predominantly level ice. A decreasing trend in sea ice thickness was observed when compared to measurements from 2003 to 2017. Notably, the spring of 2021 and summer of 2022 saw significant decreases in Sea Ice Extent (SIE). Although the overall mean sea ice thickness remained unchanged, the northwestern Weddell region experienced a noticeable decrease in ice thickness.Earth and Planetary Science
Ownership Matters: Not-for-Profit Chain Nursing Homes Have Higher Utilization of Agency Nursing Staff
No full textNursing homes (NHs) have long struggled with nurse shortages, leading to a greater reliance on agency nurses. The purpose of this study was to examine the impact of NH ownership on agency nurse utilization. Data were derived from multiple sources, including the Payroll-Based Journal and NH Five-Star Facility Quality Reporting System (n: 38,550 years: 2020-2022). A 2-part logistic regression model with 2-way fixed effects (state and year) was used to assess the association of ownership and agency nurse utilization. Model 1 compared facilities with and without agency nurse use, while Model 2 focused on NHs using agency nurses, examining high utilization (top 10%). The dependent variables were agency nurse utilization ratios for registered nurses (RNs), licensed practical nurses (LPNs), and certified nursing assistants (CNAs). The primary independent variable was ownership/chain affiliation: for-profit chain (FPC), for-profit independent (FPI), not-for-profit chain (NFPC), and not-for-profit independent (NFPI). Model 1 showed that NFPC facilities had higher odds of using agency RNs (OR = 1.65), LPNs (OR = 1.53), and CNAs (OR = 1.38) compared to NFPI facilities (all P < .001), while FPC facilities also had increased odds for RNs (OR = 1.43), LPNs (OR = 1.30), and CNAs (OR = 1.15) (all P < .001). Model 2 indicated that NFPC, FPC, and FPI facilities were more likely to be high utilizers (top 10%) of agency nurses, with NFPC facilities having the highest odds across all categories. Pairwise comparisons showed that NFPC had the highest utilization of agency RNs and LPNs compared to other ownership groups. These results highlight the significant impact of NH ownership on staffing practices, suggesting that ownership type influences agency nurse utilization.Public Healt
Optimal control theory and applications in infectious disease modeling
No full textWe explore the fundamentals of optimal control theory and the work of Lev Pontryagin's Maximum Principle [39]. The general minimization functional is defined below
[equations]
where x(t) denote the state of the system at time t, and u(t) represent the control, where t spans the interval [t<sub>0</sub>, t<sub>f</sub>]
In our first application, We use optimal control to analyze the Susceptible-Exposed-Infected-Recovered (SEIR) epidemic model, offering a comparative analysis of minimization and maximization control strategies. This approach aims to minimize the number of infected people and the overall cost of vaccines over a fixed time period T compared to maximizing the overall healthy population N(t).
Our second application focuses on two different compartmental models: the Susceptible-Vaccinated-Infected-Recovered (SVIR) and the Susceptible-Infected-Treated-Recovered (SITR). By comparing and contrasting the outcomes of vaccination and treatment as control strategies, we are able to highlight the effectiveness of each approach within the models. The comparative analysis is extended to different scenarios to simulate different stages of an epidemic.Mathematic
Measuring the Interdisciplinarity and Collaboration Perceptions of U.S. Scientists, Engineers, and Educators
No full textInterdisciplinarity has the potential to lead to more innovation and advances in knowledge than are possible from a single discipline. Yet, little is known about interdisciplinary collaborations and the perceptions of those involved. This quantitative study investigated the perceptions of U.S. faculty, staff, postdocs, and graduate students involved in education and science/engineering collaborations. Exploratory factor analysis was conducted for two modified scales, Collaboration Perceptions (CP; n = 117; 17 items; α = .923) and Interdisciplinarity Perceptions (IP; n = 119; 11 items; α = .852). Participants’ perceptions of collaboration and interdisciplinarity were strongly positive and did not significantly differ based on demographic factors (e.g., gender, discipline, role). Perceptions were influenced by collectivist orientation; the high collectivism group had significantly more positive perceptions of collaboration and interdisciplinarity, and collectivist orientation was positively and significantly correlated with CP and IP scores. Implications and recommendations for interdisciplinary collaborations will be discussed.This research was supported by the United States Department of Agriculture (award number 2017-67009-26771)
Southern Charmed: The Stigma Surrounding the Contemporary Romance Novel
No full textSouthern Charmed is a contemporary romance novel that follows the journey of hopeless romantic Cecilia Woodward who ditched her backwoods hometown of Cactus Creek to start a new life for herself. When faced with a family emergency, Cecilia is forced to come back for the first time in nearly a decade to confront the people and places she was so desperate to leave. Two of those people are a golden retriever of a law enforcement officer who keeps putting his nose where it doesn?t belong and an ex-best friend who feels betrayed that she ever left. As Cecilia reconnects with the family, friends, and flames she left behind, she finds yet another reason to hate Cactus Creek: someone has been vandalizing the family farm. Luckily, she has a friend (or enemy) who works in law enforcement. Together, the unlikely duo joins forces to catch the culprit and help restore the family?s legacy. The longer that Cece spends in Cactus Creek, the more she begins to realize that she is falling in love, something that she didn?t expect or even think possible. As she attempts to harden her heart and distance herself from anything that could tie her down, she realizes that maybe the life she tried so hard to escape isn?t so bad after all. The question remains, will Cecilia?s love for her hometown roots be rekindled, or will she escape back to her old life when given the chance?Englis
Extreme Heat and Pregnancy: A Content Analysis of Heat Health Risk Communication by US Public Health Agencies
No full textObjectives:
Exposure to extreme heat events increases the risk for negative birth outcomes, including preterm birth. This study sought to determine the presence and content of web-based heat health information for pregnant people provided by federal, state, and local government public health websites.
Methods:
This website content analysis consisted of 17 federal, 50 state, and 21 city websites, and noted which of 25 recognized pregnancy heat health data elements were included. Data for the analysis were collected from March 12, 2022, through March 16, 2022.
Results:
The search identified 17 federal websites, 38 state websites, and 19 city websites with heat health information. Within these, only seven websites listed pregnant people as a vulnerable or at-risk population, and only six websites provided information related to heat health specifically for pregnancy. Of the 25 themes recognized as important for pregnancy risk during extreme heat exposure, only 11 were represented within these 6 websites.
Conclusion:
The presence of web-based pregnancy heat health information is infrequent and limited in content. Boosting web-based publication of extreme heat and pregnancy risks could mitigate negative maternal and child health outcomes.Management, Policy and Community Healt
Deep Learning Approaches for Early Detection of Lung Cancer Using CT Scan Images
No full textLung cancer, a leading cause of cancer-related deaths worldwide, necessitates early and accurate detection to improve patient survival rates. This thesis investigates the application of machine learning models, specifically a custom Convolutional Neural Network (CNN), MobileNetV2, and ResNet50, in detecting lung cancer using CT scan images. The dataset, comprising images from the Iraq-Oncology Teaching Hospital/National Center for Cancer Diseases (IQ-OTH/NCCD), includes normal, benign, and malignant cases. The proposed custom CNN model was compared with pre-trained MobileNetV2 and ResNet50 models, with adjustments made to enhance their performance for lung cancer detection. Results indicate that ResNet50 outperformed the other models, achieving an accuracy of 84.8%, with precision rates of 90.4% for malignant, 87.3% for normal, and 64.8% for benign cases. MobileNetV2 showed an accuracy of 80.4%, with precision rates of 88.7% for malignant, 83.1% for normal, and 55.9% for benign cases. The custom CNN model achieved an accuracy of 69.4%, with precision rates of 81.5% for malignant, 73.2% for normal, and 41.1% for benign cases. The findings underscore the potential of AI, particularly ResNet50, in enhancing the accuracy and efficiency of lung cancer diagnostics. The integration of AI in lung cancer detection offers significant promise for early diagnosis and improved patient outcomes.Mechanical Engineerin
Using Project-Based Learning to Teach Water Quality and Nitrogen Cycle
No full textIntroduction:
Water quality is an essential component of the natural world and of human life. Lack of access to clean water is a global issue that affects more than 450 million children (Alhattab, 2021). An integral component of water quality is the nitrogen cycle. The complex and real world application of the nitrogen cycle and water quality naturally lends itself to STEM education (Moore & Glancy, 2020). The nitrogen cycle also incorporates biology, math, chemistry, and many other subjects. For these reasons it provides an excellent base for interdisciplinary learning/teaching.
Research Questions:
Water quality is a global issue. However, current curriculum does a poor job teaching this to K-12 students. My goal is to discover the best teaching practices to engage students in authentic learning. How can teachers utilize inquiry-based learning to help students recognize the importance of the nitrogen cycle in their lives and in real-world water quality problems? How can an interdisciplinary approach be used to effectively engage students of different ages? What are the limitations or challenges in using these methods? In summary: How can teachers engage students in project-based inquiry to increase learning and curiosity about the relationship between the nitrogen cycle and water quality?Curriculum and Instructio