UTSA Runner Research Press (Univ. of Texas at San Antonio)
Not a member yet
6846 research outputs found
Sort by
Orientation-Independent Suction Performance and Enhanced Fluid Handling in the Battlefield-Ready Innovative Suction Kit (BRISK)
No full textEffective airway management is crucial in emergency and combat scenarios, yet conventional portable suction devices are limited by orientation sensitivity – losing efficiency when tilted or inverted. In chaotic environments – such as active battlefields or during rapid patient transport – responders cannot guarantee that devices will remain upright, and traditional suction units often fail to clear obstructions if they are not correctly positioned. This limitation risks critical delays and can compromise life-saving care. The Battlefield Ready Innovative Suction Kit (BRISK) overcomes this challenge by employing an orientation-independent design that ensures reliable suction performance in any upright, tilted, or inverted position.Mechanical Engineerin
Cross-Modality Evaluation of Explainable AI Methods on Image and Audio Classification Tasks
No full textThis thesis presents an evaluation of six post-hoc XAI methods available in the Captum XAI framework for PyTorch (Saliency, Integrated Gradients, Guided GradCAM, DeepLIFT, Input × Gradient, Occlusion) applied to both image (MNIST) and audio (AudioMNIST) classification tasks using a consistent CNN architecture with grayscale inputs (MNIST Images, AudioMNIST Spectrograms) from both datasets. I benchmark these methods using faithfulness correlation and complexity metrics revealing significant modality-based performance discrepancies. I experiment with modifications to the baseline (none, median, noisy silence, and low-energy masks) input in Integrated Gradients across MNIST and AudioMNIST and observe minimal effect on performance metrics. This work contributes a reproducible, cross-modal benchmark for audio spectrogram explainability and highlights the challenges in existing interpretability techniques when transferred from image to audio modality.Computer Scienc
Multi-Scale Gaussian Smoothed Dynamic Class Activation Maps for Enhanced Visual Explanations
No full textDeep segmentation models have achieved remarkable accuracy, yet their inherent black-box nature poses challenges for interpretability. Existing methods such as Seg-Grad-CAM often lose spatial details due to global pooling, leading to noisy and less localized explanations. We introduce MSGS-CAM, a novel framework that enhances visual explanations by integrating multi-scale average pooling and Gaussian smoothing. By capturing both fine and coarse features through multi-scale fusion and reducing high-frequency noise with Gaussian filtering, MSGS-CAM produces clear and precise class activation maps. Our approach significantly improves localization accuracy and interpretability, as demonstrated by superior performance on both natural and medical image datasets.Computer Scienc
Drug Discovery and Characterization of Activity Against Coccidioides Spherules
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 August 26, 2027.Coccidioides is a genus of dimorphic fungal pathogens that cause the disease coccidioidomycosis (CM), also known as Valley Fever. The purpose of my research was to identify and characterize novel potential antifungals effective against Coccidioides, as the therapeutic options for CM are limited to only a handful of drugs in the triazole and polyene classes. My first aim encompassed screening several drug libraries, including the Broad Repurposing (5,200 compounds), Prestwick Chemicals (1,520 compounds), Selleck L8200 Antiparasitics (219 compounds), the MedChem Express CNS-Penetrants libraries (782 compounds), and the compound niclosamide ethanolamine (Adipogen), totaling 7,722 compounds screened. Libraries were screened using XTT assays to find hits that inhibited spherule initial growth more than 70% compared to DMSO controls and had B-score <-3 when normalized across all plates screened. Compounds were parsed and removed if they had previously been designated as toxic or as antifungals. Twenty-seven compounds met our initial cutoffs, and three additional compounds from our collaborators were also investigated to characterize their activity against the parasitic cells (spherules) of Coccidioides.
We characterized these 30 drugs using several methods. We profiled them for their antifungal activity (IC50) and cytotoxicity (CC50) in a HepG2 model using microdilution assays. We combined the compounds with amphotericin B (AmB), fluconazole (FLU), or caspofungin (CAS) to determine if the compounds acted synergistically when combined with clinical antifungals. Five compounds, including 10058-F4, niclosamide (NIC), niclosamide ethanolamine (NEN), oxethazaine (OXE), and pentamidine isethionate (PENT-I) demonstrated synergy with AmB. This discovery may facilitate novel therapeutic strategies and lower the required dose of AmB, which is highly toxic. These five drugs were tested in vivo using a Galleria mellonella model of CM to measure the survival and fungal burden of larvae following treatment. To better elucidate the mechanisms through which these five drugs acted, they were compared structurally to each other and other clinical antifungals, examined microscopically, and screened for location-specific staining using image flow cytometry. Spherules were also subjected to RNA-seq to identify their potential impacts and targets within known antifungal pathways and from their unique signatures. qRT-PCR further validated the RNA-seq data to confirm the drug impacts on Coccidioides growth.Molecular Microbiology and Immunolog
Evaluating Stormwater Filtration Basin Performance in the Edwards Aquifer Recharge Zone Using Microbial Source Tracking
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 August 26, 2030.Best management practices (BMPs) have been utilized in recent times to minimize peak flow and attenuate contamination during stormflow conditions. While a plethora of BMP types exist and have been researched extensively by means of examining their performance in reducing a variety of contaminant concentrations, fewer studies have focused unilaterally on stormwater detention basins – specifically, their ability to reduce microbial contaminant concentrations remains a relatively novel research focus that necessitates further analysis. BMPs are of particular importance in a large portion of land in and around San Antonio, due to the wide-spanning area of the Edwards Aquifer recharge zone. Karst structures all across the world – including the Edwards Aquifer in San Antonio – are critical sources of drinking water, and their recharge areas possess a unique vulnerability not shared by the vast majority of other aquifer types. As a result, preserving and ensuring the potability and overall caliber of water quality is vastly important.
Stormwater detention basins, like other BMP types, have the capability of mitigating runoff contamination – the question remains, which pollutant types are the most significant in their reduction with such structures. The current study seeks to evaluate and quantify the degree to which two stormwater detention basins (TPC basin, Kyle Seale basin), located overlying the Edwards Aquifer recharge zone, are able to attenuate and reduce microbial contamination in stormwater runoff during stormflow conditions. The current study builds off a past study (Moghadam et al., 2023) which examined the nutrient removal capabilities of the same two basins by (1) examining the degree to which the basins were effective in removing key microbial source tracking (MST) markers (HF183, Humanmt-ND5, BacCan, Entero1a) and general fecal coliforms, and (2) assessing the degree to which MST markers correlated with the nutrient removal efficacy found in Moghadam et al.’s 2023 analysis.
The current study found limited marker removal efficacy and significant fecal coliform removal efficacy in TPC basin and negligible marker removal efficacy and limited fecal coliform removal efficacy in Kyle Seale basin – likely a result of the significantly larger total area of TPC basin than the latter, due to the increased sedimentation efficiency provided by the additional area in TPC basin. Additionally, the study found notable correlation between the marker Humanmt-ND5 and nutrient removal capabilities in TPC basin, in addition to notable correlation between fecal coliform concentration and TSS levels in TPC basin, though these trends were absent in Kyle Seale basin. The current study recommends future research to be conducted utilizing a larger quantity of basins to be sampled over a longer duration to capture a larger sample size of stormflow events, in order to assess the degree to which the correlations found in the current study are able to be maintained over more significant test sizes.Civil and Environmental Engineerin
Leveraging team familiarity to improve provider retention and OR efficiency
Full text linkThis poster was presented at the 2025 Postdoctoral Appreciation Week event.Operating rooms (ORs) are among the most resource-intensive hospital units, costing an estimated 40,000–$80,000 per provider and vacancies often taking more than six months to fill. Addressing both efficiency and retention requires strategies that improve working conditions while optimizing resource use. One promising approach is team familiarity—ensuring that providers consistently work with trusted colleagues. Familiarity has been shown to enhance collaboration, reduce errors, and improve job satisfaction, yet it is rarely embedded into scheduling or predictive analytics. To evaluate its impact, we analyzed historical shift data from a Level I trauma center and case data from a large pediatric academic medical center. A hybrid recommender model (LightFM) was developed to capture anesthesiology providers’ scheduling preferences, achieving strong predictive accuracy (AUC = 0.84). These insights form the basis of an optimization framework that balances individual preferences with concurrency and coverage requirements, enabling fair and engagement-driven shift assignments. In parallel, we assessed the effect of familiarity on surgical duration prediction. Using machine learning methods—including LASSO, Random Forest, and LightGBM—we found that LightGBM achieved the best performance, reducing median prediction error from –12.00 to –0.69 minutes. Notably, team familiarity emerged as one of the top predictors of case duration, underscoring its role in shaping operating room performance. By integrating preference modeling, predictive analytics, and optimization, this work demonstrates how team familiarity can serve as a unifying principle to improve both provider retention and operating room efficiency. Because the models draw on existing scheduling and electronic health record data, implementation feasibility is high, requiring minimal additional data collection. These findings suggest that data-driven scheduling systems grounded in team familiarity offer a path toward more sustainable workforce practices and more efficient pediatric surgical care.Operations and Analytic
Nitrate Monitoring in Semi-Urban Groundwater of Northeastern Saudi Arabia
Full text linkMonitoring nitrate levels in water is critical to protect public health and ensure compliance with regulatory standards. This study provides a comprehensive evaluation of four analytical techniques&mdash;test strips, ion-selective electrodes (ISE), colorimetric methods, and titration&mdash;to assess nitrate levels in a variety of water sources, including standard solutions, rainwater, bottled water, and groundwater from both shallow and deep wells located in semi-urban regions of Saudi Arabia. Each method was assessed for sensitivity, accuracy, detection limits, reproducibility, and operational practicality. Test strips offer rapid, low-cost screening but consistently underestimate nitrate concentrations, particularly at low levels. The ISE demonstrated broad applicability and reliable performance across a wide concentration range when properly calibrated, making it suitable for both field and laboratory applications. Colorimetric methods provide excellent sensitivity for trace-level detection, whereas titration delivers the highest accuracy for high-nitrate samples despite its time-intensive nature. By calibrating and validating the methods against certified standards, we quantitatively demonstrated their reliability through statistical measures such as precision and accuracy rates. Moreover, the application of Geographic Information System (GIS) techniques in spatial analysis has revealed significant differences in the distribution of nitrates. Notably, shallow wells located in the northern regions surpass the 50 mg/L limit set by the World Health Organization (WHO), thereby indicating the presence of localized contamination hotspots. This study is among the first to systematically compare nitrate detection methods across a wide range of water types in a semi-urban area of Saudi Arabia. Building on a detailed analysis of each method, we underline the crucial need for the strategic selection of nitrate analysis techniques. This selection should be tailored to specific operational contexts, accuracy requirements, and concentration ranges to guide stakeholders towards more informed decision-making. These findings provide actionable guidance for public health officials and water managers to prioritize monitoring, safeguard drinking-water sources, and mitigate nitrate-related health risks in semi-urban communities
Existential Purgatory: Colonial Ideology & a Response Through Indigenous Philosophy Pathways
Full text linkThis thesis aims to uncover the governing ideology within the United States which dictates our societal existence and praxis, thereby analyzing how false consciousness has permeated recent American history. Ideology is conceptualized here as a historical idealism concretized into society through a process of totalization, arguing that specific conceptual structures were deployed to benefit particular demographics. By examining how ideology is ingrained in societal praxis, we can uncover the underlying function of pervasive behaviors and beliefs. To further understand these functions and false consciousness, this work engages with the writings of Frantz Fanon and Jean-Paul Sartre. The analysis of Fanon and Sartre begins by highlighting a philosophical tension, leading to a reconciliation that defines the thesis’s central concept: Existential Purgatory. This concept names the state of being that follows separation from dominant ideology and false consciousness—a profound self-exile.
While this self-exile and "loss of everything" can be frightening, I argue that Indigenous philosophies, particularly the works of Jessica Hernandez and Brian Burkart, offer a path forward: the creation of a new, reciprocal relationship with the world. This relationship necessitates moving beyond an anthropocentric worldview to fully recognize and embrace the sacredness of all life.Philosoph
Polyelectrolyte Properties of Actin and Microtubule and Their Roles in Electrical Activities in Neurons
Full text linkThe electrical activity of neurons underlies nearly all functions of the nervous system. Neurons transmit information as electrical impulses, which are crucial for coordinating behavior, thoughts, sensation, and movement. One of the most significant breakthroughs in neuroscience is the Hodgkin-Huxley cable theory, which explains the generation of the action potential. Introduced over 70 years ago, this theory presented a conductance-based transmission-line model to explain how voltage-dependent ion channels generate and propagate electrical impulses, a discovery recognized with the 1963 Nobel Prize in Physiology or Medicine. However, it remains unclear how electrical signals transmit information between the cell membrane and the nucleus, which are interconnected via an intracellular network of actin filaments and microtubules. Recent experiments suggest that these cytoskeletal filaments can conduct ionic currents, amplify signals, and generate electrical oscillations. This dissertation investigates their ability to transmit electrical signals at the nanoscale through localized ionic wave packets and examines their potential role in intracellular signaling and information processing. Although experimental evidence for ionic conduction and oscillation in cytoskeletal filaments continues to accumulate, the underlying electrodynamic mechanisms remain poorly understood. Existing theoretical models often overlook critical features, and many computational approaches are either analytically restrictive or computationally expensive. To address these challenges, we developed a quantitative and efficient framework for cytoskeletal filaments that links molecular conductance to electrical signaling in excitable cells. The framework is based on advanced ionic-conductance transmission-line models that incorporate atomistic details and biological environments to describe the electrodynamic behavior of actin filaments and microtubules under both physiological and pathological conditions. For F-actins, we formulated and implemented a single-transmission-line model to analyze ionic wave propagation under varying temperatures, pH levels, and structural configurations, as well as to examine the electrostatic effects of disease-linked mutations. For microtubules, we developed a two-coupled-transmission-line model in which the outer and luminal ionic layers along the filament are dynamically interconnected via transistor-like, voltage-dependent nanopores rather than ion channels. The model reveals that the alternating transfer of electrical energy through consecutive nanopores results in oscillations as outer and inner ionic wave packets periodically overtake one another, enabling long-distance electrical signal transmission with minimal power loss. It predicts an oscillation frequency of approximately 39 Hz, consistent with experimental findings on microtubules and notably within the gamma-band range of brain activity. In parallel, we analyzed dynamic light scattering measurements of microtubule fluctuations using an optimized scattering theory to reconstruct decay-rate and contour-length distributions from autocorrelation data. This approach establishes a quantitative connection between filament structure and its electromechanical properties. We encoded our computational approaches and theories for signal propagation along F-actins and microtubules into user-friendly, open-source applications to support experts and non-experts. Collectively, this research establishes that the cytoskeleton can be conceptualized as a nanoscale analog of neuronal electrical networks, providing new insights into intracellular communication, biocomputation, and the biophysical foundations of neurophysiology and disease.Physics and Astronom
Satisfied Enough to Take Action? The Role of Neighborhood Perceptions on Disaster Preparedness Behaviors in the United States
Full text linkThis study examines how neighborhood perceptions, measured through satisfaction and observable conditions, relate to risk perception and shape residents’ disaster preparedness behaviors. It employs regression models using the 2017 American Housing Survey data. Findings indicate that households satisfied with their neighborhoods are more likely to engage in disaster preparedness behaviors. Moreover, the presence of abandoned structures or the lack of good schools discourages such actions. In communities with low awareness of disaster risks, improving neighborhood conditions can encourage disaster preparedness behaviors and increase community protection against disaster risks.Architecture and Plannin