Doctoral Dissertations

Permanent URI for this collection

https://hdl.handle.net/10919/11041

Browse

Now showing 1 - 20 of 18567

Multi-Agent Integer Programming Games with Non-Cooperation and Cooperation
Lee, Hyunwoo (Virginia Tech, 2026-09-03)
This dissertation studies integer programming games (IPGs), in which each of several self-interested players solves an integer program whose payoff is coupled to the others' discrete decisions. Because feasible sets are combinatorial and best responses require integer optimization, even deciding whether a pure Nash equilibrium (PNE) exists is computationally intractable in general. The unifying thesis is that the equilibrium of such a game is a computable and leverageable object around which better system-level decisions can be built, and this idea is developed across the non-cooperative, generalized, and cooperative regimes on a common branch-and-cut foundation of globally valid inequalities and best-response oracles. First, to make equilibrium computation scale, round-based random-restart best-response dynamics (RRR-BRD) are introduced with a Las Vegas convergence guarantee and hybridized with the zero-regret cutting-plane method (BZR), lifting tractable PNE computation and sampling from a handful of players to dozens; the methods are instantiated on edge-weighted budgeted maximum coverage games, including a Minnesota aquatic invasive species (AIS) inspection game with up to 84 counties. Second, Nash-informed planning converts computed equilibria into individual-rationality floors and optimizes social welfare above them, improving system outcomes without making any player worse off; the accompanying Price of Individual Rationality is proven to be bounded by the Price of Stability, and the welfare cost of guaranteeing participation is empirically negligible. Third, for generalized IPGs with shared constraints, conditional equilibrium inequalities and a generalized zero-regret algorithm optimize over the equilibrium set, while α-approximate equilibrium certificates and structural (VEST) cuts handle cost-sharing games such as integer splittable bin packing. Fourth, cooperative IPGs define coalition values through pooled integer programs and compute optimal, Core-stable coalition structures via stability inequalities and separation. Computational studies, anchored by the AIS application, demonstrate scalability, interpretability, and concrete policy improvements.
Partial resistance to Schistosoma mansoni in mice inoculated with cercarial bodies
Mecham, John Alvin (Virginia Tech, 1972-07)
One group of mice was inoculated with Schistosorna mansoni cercarial bodies. Another group was inoculated with cercarial tails and a third group inoculated with cercarial bodies plus tails in an attempt to determine if schistosome cercarial parts elicit differing immune responses. The mice were then challenged with 100 viable cercariae as were control mice. At six weeks after infection, adult schistosomes were recovered by perfusion. The average worm burden for mice inoculated with cercarial bodies was 9.07 compared to 21.00 for mice inoculated with cercarial tails, 21.90 for mice inoculated with bodies plus tails, and 23.30 for control mice. The number of adults recovered from mice inoculated with cercarial bodies was significantly less than from mice inoculated with cercarial tails, or bodies plus tails, and control mice. Two serodiagnosic tests, Cercarien-hullen Reaction (CHR) and immunodiffusion, were conducted with sera from experimental and control mice. CHR tests were positive for all experimental mice with strongest reactions occurring with infected mice and with mice inoculated with cercarial bodies. Precipitating antibodies against whole cercarial antigen were shown using anti-body, anti-body-plus-tail, and infected mouse serum. No precipitating antibodies were shown against whole cercarial antigen using anti-tail mouse serum. The possibility of a cercarial body-tail interaction resulting in stimulation to develop or protection for the developing schistosomulum is indicated.
Self-Reported ADHD Symptoms and Intraindividual Variability in Momentary Cognition Among Young Adults
Mansoor, Marrium (Virginia Tech, 2026-08-31)
Previous research examining intraindividual variability (IIV) on cognitive tasks among individuals with ADHD has tended to focus on variability in reaction times on cognitive task trials. However, intraindividual variability that occurs in moment-to-moment daily cognition has not been examined thus far. This study utilized an Ecological Momentary Assessment (EMA) methodology to investigate within-individual variability across measurement occasions and its association with self-reported ADHD symptoms as well as daily lifestyle activities. Data was collected from 116 young adults (M age = 19.9 years, 68% female, and 44% White) who completed a baseline session in the lab, followed by three semi-randomized EMA sessions per day over two weeks. Each EMA session included three cognitive tasks and a counterbalanced selection of additional surveys. Data was analyzed using Mixed Effects Location Scale models. Results indicated that higher ADHD symptoms were associated with greater variability in reaction times for tasks requiring cognitive flexibility and working memory, but less variability in reaction times on an inhibitory control task. Several interaction models were then conducted to investigate if daily activities influenced the association between self-reported ADHD symptoms and cognitive variability. Several of these were found to be significant, however there was considerable heterogeneity in the results. This study is the first to examine intraindividual variability across occasions in the context of ADHD symptomology and represent an important step towards better understanding of daily cognition in young adults with ADHD symptoms.
Toward Trustworthy Health AI Systems: Advancing Clinician–AI Interaction Through Interpretability, Fairness, and Context-Aware Multi-Agent Safety Architectures
Nasarian, Elham (Virginia Tech, 2026-08-31)
AI has become increasingly integrated into healthcare, supporting clinical decision making, patient risk prediction, and patient-facing information systems. Despite substantial advances in predictive modeling and LLMs, widespread adoption of AI in healthcare remains constrained by challenges related to interpretability, fairness, safety, and human- AI interaction. Healthcare applications require not only accurate predictions and recommendations but also transparent, equitable, and clinically reliable systems that can be trusted by end-users. This dissertation advances the design of trustworthy health AI systems through three complementary studies focused on clinician–AI interaction, fairness-aware clinical risk prediction, and context-aware multi-agent safety architectures. In essay 1, a systematic review of explainable AI (XAI) in healthcare synthesizes existing approaches and proposes a clinician-centered framework for improving collaboration between AI systems and healthcare professionals. The review identifies key challenges in implementing interpretable AI within clinical decision support systems and provides a roadmap for responsible deployment. In the second essay, a fair and clinically interpretable machine learning framework is developed to predict distinct opioid-related respiratory deterioration events among hospitalized patients receiving opioid therapy. Using electronic health record (HER) data, the proposed multiclass framework differentiates Naloxone intervention events, Blue Code respiratory arrest events, and Rapid Response events. The framework integrates explainability methods, fairness evaluation, and age-specific threshold optimization to improve detection of severe respiratory outcomes while maintaining clinical interpretability. Experimental results demonstrate substantial performance improvements compared with traditional clinical risk scoring approaches. The third essay, a context-aware multi-agent safety architecture, CareGuardAI, is introduced to address clinical safety risks and hallucination risks in patient-facing medical LLMs. The proposed system combines safety-constrained generation, risk assessment agents, and iterative refinement mechanisms to evaluate both medical safety and factual reliability before responses are delivered to patients. Across multiple healthcare safety and hallucination benchmarks, the architecture demonstrates improved performance relative to strong baseline models while maintaining bounded response latency. Collectively, these studies contribute a unified perspective on trustworthy health AI by integrating interpretability, fairness, and safety into the design of healthcare AI systems. The findings provide practical and methodological guidance for developing AI-enabled clinical decision support systems that improve effective human–AI interaction (clinicians and patients) and responsible deployment in high-stakes healthcare environments.
WaveHoltz: Theory and High Performance Applications
Rotem, Amit (Virginia Tech, 2026-08-28)
The solution to the Helmholtz equation describes the spatial waveform of a time-harmonic wave and arises in acoustics, electromagnetism, and seismology. Its efficient numerical treatment makes engineering problems tractable but presents two central difficulties. First, the waves must be sufficiently resolved. Engineering problems often span hundreds to thousands of wavelengths, over which phase error accumulates, demanding many points per wavelength and yielding systems with millions to billions of degrees of freedom. Second, standard discretizations produce a complex-valued, highly indefinite linear system on which general-purpose iterative methods perform poorly or fail outright, making specialized methods for the equation a challenging, active research topic. This dissertation advances one such method: WaveHoltz, which relates the Helmholtz equation to its time-dependent counterpart, the wave equation. WaveHoltz evolves the wave equation and filters its solution over a short time interval to estimate the Helmholtz solution; this estimate is fed back as an initial condition, iteratively improving the estimate until convergence; this iteration can be accelerated by standard Krylov methods. On the theoretical side, convergence of the discrete iteration is established, the temporal error of the wave solve is eliminated, and WaveHoltz-HMM, a homogenization method with error estimates, is introduced. On the computational side, a high-performance GPU implementation combining multigrid-accelerated time stepping with massively parallel algorithms and a GPU-tailored domain decomposition approach made efficient by using WaveHoltz as a subdomain solver are presented. Altogether, WaveHoltz proves robust and efficient for solving the Helmholtz equation.
Assessment of Peanut (Arachis hypogaea L.) Responses to Heat and Drought Stress Through Physiological, Machine Learning, and Transcriptomic Approaches
Vennam, Ranadheer Reddy (Virginia Tech, 2026-08-28)
Heat-exacerbated drought stress is an increasing constraint on peanut (Arachis hypogaea L.) production in the rainfed Virginia–Carolina region, yet the physiological, phenotypic, and molecular consequences of this combined stress remain incompletely characterized. This dissertation integrated growth chamber, multi-year field, high-throughput phenotyping, and transcriptomic approaches across five interconnected studies to address this gap. In growth chamber and field trials, combined heat and drought stress (HS+DS) significantly reduced stomatal conductance and transpiration relative to control or rainfed treatment. Under sustained mid- to late-season HS+DS imposed with rainout shelters across two field seasons, canopy temperature increased by up to 6 °C, accompanied by greater leaf wilting severity and pod yield reductions of 52–73%. Yield, grade, and economic analyses across 12 genotypes showed that HS+DS reduced the percentages of extra-large kernels and total sound mature kernels and eliminated profitability, with losses exceeding $2,000 ha⁻¹ in 2025. However, breeding line 14x009-1-5-1-1 maintained comparatively stable pod yield across both years. Leaf wilting and stomatal conductance were the strongest predictors of yield loss (R² = 0.86–0.89). To scale wilting assessment beyond manual scoring, a machine learning framework combining UAV multispectral imagery and accumulated growing degree days was developed utilizing data across three years and five sites; Extreme Gradient Boosting achieved the highest predictive accuracy (R² = 0.82–0.87), enabling pixel-level mapping of wilting severity. Finally, RNA sequencing of four virginia-type genotypes under field HS+DS identified 1,733 differentially expressed genes, revealing that genotypes with similar physiological responses employed largely distinct transcriptional strategies. Collectively, these findings demonstrate that HS+DS imposes convergent physiological and yield penalties on peanut via distinct molecular pathways, while also establishing scalable phenotyping tools that can support management and breeding strategies for improved resilience to heat and drought in rainfed peanut production.
Navigating Complexity in Agricultural Transitions through Living Laboratory: Knowledge Co-production, Practice Transformation, and Precision Agriculture Adoption
Joshi, Bhavna (Virginia Tech, 2026-08-27)
This dissertation examines how precision agriculture knowledge is produced, how it shapes farmer perceptions and practices, and how it informs adoption decisions among the crop advisors through a Complex Adaptive Systems framework. This study employs both secondary data obtained from an NSF-funded living laboratory project situated in South Dakota, Virginia, and Vermont, as well as primary data. The first chapter examined knowledge-production within living laboratories from the research team's perspective. It found that knowledge was produced through negotiation rather than integration, and that structural misalignment between the living lab's transdisciplinary aspirations and the institutional architecture of federally funded research constrained genuine co-production, while still producing real epistemic transformations among team members. The second study examined what participation produced for farmers, adapting Geenhuizen's evaluation framework to account for farmers' inputs and exogenous influences. It found modest, unevenly distributed ripple effects shaped more by farmers' pre-existing orientations than by project design. A federal funding disruption triggered differential engagement among full-time commercial farmers, who were dependent on the funds and subsidies for their livelihood. The third study examined AI adoption among crop advisors using an adapted C-TAM-TPB framework. Perceived usefulness was the strongest predictor of adoption. Data concerns showed a dual role, positively predicting adoption and reflecting anticipatory engagement rather than resistance.
From Margin Calls to Micro-Expressions: Leverage, Liquidity, and Physiological Signatures of Experimental Bubbles
Spoon, Ross Linwood (Virginia Tech, 2026-08-26)
This dissertation develops new experimental infrastructure to study instability in speculative asset markets, focusing on short squeezes and bubble-crash cycles alongside the behavioral and physiological states that accompany them. The first chapter examines short squeezes: rapid and extreme price increases caused when speculative traders who borrowed an asset are forced into the market in order to acquire it again. This phenomenon is largely absent from experimental asset-pricing research despite its real-world prominence. By permitting leveraged short positions and enforcing margin requirements, the experimental design generates conditions under which squeezes can poten- tially emerge. We develop a tiered classification of squeeze events that includes voluntarily buying the asset at a loss and being forced to purchase the asset to maintain margin re- quirements; in the most extreme cases, cascading buy-ins drove prices to nearly 50 times fundamental value. We conclude that short-selling restrictions and excess liquidity amplify bubble formation, which creates an environment in which these explosive squeezes are more likely to happen. The second chapter turns to the psychological and physiological underpinnings of bubble formation and collapse in constant fundamental value markets. Using Support Vector Re- gression alongside linear models, we find that conventional financial signals, such as lagged returns and order book imbalances, outperform biometric measures in predicting contempo- raneous market returns. At the individual level, however, electrodermal activity meaningfully predicts de-risking behavior and revealed risk aversion, with high earners exhibiting antici- patory arousal before crashes and low earners showing only reactionary spikes. The third chapter introduces oPAL (open-source Periodic Auction Laboratory), a turnkey, extensible asset market platform built on the oTree architecture. oPAL fills a critical gap left by existing tools' near-exclusive focus on continuous double auctions, offering discrete call markets that isolate strategic bidding from execution-speed noise. Additionally, oPAL addresses practical frictions of online synchronous recruitment through an integrated pre- screening module and administrative portal. Key features enabling the two experiments presented in this dissertation include short selling, margin-based leverage, automated liquidations, flexible fundamental value trajectories, and a motion-artifact-minimizing entry interface for biometric recording. Together, these chapters contribute new open-source infrastructure for experimental finance and new evidence on how leverage, liquidity, and emotional state jointly shape the formation and resolution of price instability in speculative markets.
Application-Aware I/O Prediction and Optimization for HPC and AI Workloads
Yazdani, Ahmad Hossein (Virginia Tech, 2026-08-26)
High-performance computing (HPC) systems are essential for executing large-scale scien- tific and artificial intelligence workloads that demand extensive parallelism across compute, storage, and memory resources. Modern systems, such as Aurora at Argonne National Lab- oratory and the exascale Frontier system at Oak Ridge National Laboratory, have significantly expanded computational and I/O capabilities. However, storage and memory subsystems often struggle to keep pace with the scale, diversity, and dynamic behavior of modern applications. This gap creates performance bottlenecks that are difficult to diag- nose and mitigate using rigid, application-agnostic I/O management strategies. Addressing these challenges requires a deeper understanding of how application semantics, execution configurations, and workload behavior interact with storage and memory systems. This thesis addresses these challenges across three complementary directions. First, we model application I/O behavior using historical execution patterns and user-specified parameters. Through analysis of E3SM and LAMMPS workloads, we show that prior application con- figurations are strongly correlated with future I/O behavior, enabling accurate prediction over 1–10-day windows with uncertainty below 0.25 and 80–90% prediction accuracy. Sec- ond, we introduce IO-Sense, an agentic I/O reasoning pipeline that combines Darshan logs with application-level configuration semantics to summarize large-scale traces and explain how configuration knobs influence I/O patterns. On AMReX-based HPC applications, IO- Sense achieves over 90% log similarity with the ground-truth log derived from the knob change for most evaluated models and demonstrates that finer-grained I/O counters can improve reasoning accuracy, though at the cost of higher inference overhead. Third, we study I/O and memory behavior in deep learning training workloads through SHADE and SODA-LLM. SHADE introduces a rank-based importance sampling policy that improves data locality and cache efficiency for large-scale computer vision training workloads. Build- ing on this direction, SODA-LLM targets transformer-based language model training/fine- tuning by exploiting the spatial and temporal stability of activation importance to guide token-level activation retention, offloading, and prefetching during the backward pass. The proposed system reduces end-to-end training latency by up to 20% while lowering memory usage by 5%. Together, these projects show that application-aware reasoning can improve prediction, diagnosis, and optimization across HPC and AI workloads. By connecting appli- cation configurations and execution semantics to observed I/O and memory behavior, this thesis provides a foundation for more adaptive storage and memory management strategies in next-generation computing systems.
Understanding and Enhancing Sequential Decision-Making and Alignment in Foundation Models
Sel, Bilgehan (Virginia Tech, 2026-08-26)
Foundation models are used as sequential decision-makers, in tasks whose outputs unfold as dependent steps: reasoning, planning, decisions affecting several parties, and generation under safety requirements. This dissertation asks which factors govern the performance of foundation models in such settings and how that performance can be improved. The answer locates those factors in training data and training procedure. Pretraining text preserves finished solutions far more often than the failed attempts behind them, so a model trained on it defaults to confident, linearly coherent continuation; exploring alternative paths and retracting mistaken steps are underrepresented behaviors, not absent capabilities. The contributions elicit these behaviors in context and then teach them by supervised and reinforcement learning. One observation recurs: a model given room to explore diverse solution paths and to back out of its own mistakes makes better decisions, and in safety-critical settings the same capacity lets it recover from unsafe trajectories. The body has three parts. Part I treats foundation models in reinforcement learning: a meta-learning method for sequences of derivative-free optimization tasks, with task-averaged regret guarantees; policy optimization under several reward objectives and hard safety constraints, with a rectification step that restores feasibility after a detected violation; and an analysis tracing in-context reinforcement learning to the diversity of the pretraining task distribution. Part II treats large language models in sequential decision-making: a tool-use framework that translates natural-language energy-management requests into solver-ready optimization programs; the Algorithm of Thoughts, a prompting strategy whose exemplars record a search process so that the model explores, prunes, and backtracks within a single generation; an extension to autonomous long-horizon planning; and a training pipeline that makes the behavior a concise default. Part III turns the same capacity to alignment and safety: a prompting framework that surveys a decision's consequences for every affected stakeholder before answering, and a reinforcement-learning method that trains the backtracking step as a safety signal, so that the model retracts an emerging violation and continues from the safe prefix. Recovery complements avoidance rather than replacing it; an integrated red-team study of a static classifier defense motivates judging safety on the generated trajectory.
Epidemic Modeling with Generative Agents: Methodology, Prompt Sensitivity, and LLM Sensitivity
Williams, Ross Fantus (Virginia Tech, 2026-08-26)
Generative agent-based modeling (GABM) replaces the rule-based decision-making of classical agent-based models with agent decisions powered by large language models (LLMs), enabling researchers to simulate human behavior that resists clean mathematical specification. This dissertation establishes the methodology, characterizes its sensitivity to prompt design, and characterizes its sensitivity to the underlying LLM. Three studies — conducted within a shared epidemic GABM framework in which generative agents decide each day whether to stay home or go out — support the central claim: GABM is a viable simulation method, but the prompt and the LLM are both parameters of any GABM finding. Agents endogenously self-isolate as community case counts rise and quarantine when they feel ill. Collectively, this produces multi-wave dynamics followed by an endemic period. Heterogeneous behavior emerges from Big Five personality traits, age, and gender. The prompt is a parameter of GABM, but only along certain dimensions. Synonymous rewording and persona names leave epidemic trajectories statistically unchanged; minor wording variations and contextual shifts both alter outcomes. LLM choice is also a parameter. Across 21 LLM configurations from OpenAI, Anthropic, and Google, most LLMs agree on the direction of persona effects but disagree several-fold on magnitude. Standard model attributes — size, version, knowledge cutoff, release date — do not predict where an LLM's collective agent behavior lands. Mentioning a trait in the response text moves that trait's effect on the decision beyond the trait alone. Together the three studies argue that GABM is most defensible for qualitative claims about agent behavior and weakest for the quantitative claims modelers most often want to make. LLM sensitivity testing should become standard practice in GABM publications, alongside prompt sensitivity. The methodology is real; its parameters include the prompt and the LLM.
Challenges, Solutions, and Opportunities for Cyber Security in 5G Enabled Networks
Dessources, Dimitri Achilles (Virginia Tech, 2026-08-26)
The swift transition toward 5G Standalone (SA) and Open Radio Access Network (O-RAN) architectures represents a paradigm shift toward highly flexible, software-defined telecommu- nications infrastructure. While these advancements offer transformative benefits in multi- vendor interoperability and network agility, they also significantly increase the threat land- scape, exposing critical vulnerabilities that have persisted across several cellular generations. This dissertation establishes a holistic security framework for 5G-enabled networks. First, a comprehensive survey of Fifth Generation (5G) SA signaling security is conducted, system- atically cataloging Access Stratum (AS) and Non-Access Stratum (NAS) messages transmit- ted in clear text during the pre-authentication window of the registration procedure. The analysis identifies critical information elements, including the 5GS Mobile Identity, User Equipment (UE) Security Capability, and Radio Resource Control (RRC) configuration pa- rameters, that remain exposed to interception and manipulation by adversaries operating Rogue Base Stations (RBSs). Second, a structured threat modeling and governance frame- work for O-RAN is proposed, informed by the MITRE FiGHT matrix and O-RAN Alliance Working Group (WG)11 security principles. Six critical threat categories are characterized, namely supply chain compromise, Fifth Generation NodeB (gNB) component compromise, rogue xApp/rApp exploitation, network sniffing and spoofing, Continuous Integration/Con- tinuous Delivery (CI/CD) pipeline exploitation, and lateral movement via container escape. A Responsible, Accountable, Supportive, Consulted, Informed (RASCI) responsibility matrix is developed to delineate security accountability across network operators, vendors, standards bodies, and testing facilities throughout the O-RAN security lifecycle. Third, a fuzz testing framework for automated 5G O-RAN vulnerability assessment is demonstrated through two complementary tools. O-FUEzzer embeds Abstract Syntax Notation One (ASN.1)-aware fuzzing directly within the OpenAirInterface5G (OAI5G) UE protocol stack to target RRC message fields, while a modified 5GReplay tool performs mutation-based fuzzing of Open Fronthaul (O-FH) interface traffic. Experimental results on an O-RAN laboratory testbed reveal that fuzzing specific O-FH fields induces complete network failure with outages exceed- ing 60 seconds, confirming the feasibility and necessity of automated, field-specific vulnera- bility testing suitable for integration into CI/CD pipelines and Open Testing and Integration Centre (OTIC) certification regimes. Fourth, a lightweight Ed25519 digital signature-based one-way authentication framework is proposed to enable UE verification of gNB identity before initiating the Random Access Procedure (RAP), directly mitigating the RBS threat. The framework embeds a 104-byte authentication payload, comprising an Ed25519 signa- ture and X.509 certificate fingerprint, within the System Information Block (SIB)1 lateNon- CriticalExtension field and incorporates timestamp-based replay protection. Experimental evaluation demonstrates 100% RBS detection across all tested scenarios, an end-to-end ver- ification latency of 3.4 ms (2.2% of the SIB1 broadcast period), full backward compatibility with legacy UE deployments, and no degradation to SIB1 broadcast periodicity. Fifth, the security analysis is extended to 5G Non-Terrestrial Networks (NTNs) by examining SIB 19, the broadcast message introduced in Third Generation Partnership Project (3GPP) Re- lease 17 that provides NTN-specific configuration and satellite ephemeris data essential for UE linkability to NTN gNBs. The threats and vulnerabilities inherent to SIB 19, includ- ing spoofing of ephemeris parameters, manipulation of NTN timing advance information, and exploitation of unprotected satellite linkability data, are systematically analyzed. The Ed25519 one-way authentication framework is then adapted to secure SIB 19, enabling UE verification of NTN gNB authenticity prior to initial access in satellite and High-Altitude Platform Station (HAPS) scenarios. Preliminary results demonstrate the viability of this extension under the unique constraints of NTN environments, including longer propagation delays and ephemeral cell visibility. Together, these contributions form a cohesive security framework that identifies vulnerabilities in 5G signaling, contextualizes them within a multi- stakeholder governance model, automates their discovery through fuzz testing, mitigates a critical class of attacks through pre-authentication base station verification, and extends these protections to non-terrestrial platforms.
Characterizing Distal Tibia Fractures in Motor Vehicle Crashes: Assessing Occupant Risk, Vehicle Crashworthiness, and Clinical Outcomes
Bangert, Laurence Garrett (Virginia Tech, 2026-08-25)
High-energy distal tibia fractures are a clinically complex injury associated with high rates of clinical complications, chronic pain, and functional limitations. One of the primary causes of severe distal tibia fractures are motor vehicle crashes. Road traffic injuries account for a significant proportion of the global burden of traumatic injuries; accordingly, a large and diverse community has formed with the goal of reducing serious crash injuries to zero. Distal tibia fractures caused by motor vehicle crashes are associated with complications in 20-50% of cases, and high-energy fractures often result in the development of post-traumatic osteoarthritis, costing millions and significantly impacting the quality of life of tens of thou sands of crash occupants in the United States annually. Despite the link between high-energy distal tibia fractures and motor vehicle crashes, a gap in knowledge exists in the current lit erature regarding the relationship between severe fracture loading mechanisms and occupant demographics, vehicle crashworthiness, and crash severity. This knowledge gap limits the capacity of vehicle safety systems to prevent loading conditions necessary for severe distal tibia fractures and hinders progress towards the goal of zero serious road traffic injuries. To address this gap, the goal of this dissertation was to characterize distal tibia fracture incidence and severity in real-world crash data using established statistical designs, novel machine learning algorithms, and physics-based computational modeling. A multivariate logistic regression was trained using data from the Crash Investigation Sampling System to identify distal tibia risk factors in a large-scale crash population. Next, a machine learning classifier was trained using data from the Crash Injury Research and Engineering Network to predict clinical fracture types following a motor vehicle crash without radiology. A SHapley Additive exPlanation interpretability test was used to explain the underlying relationships between predictions of the "black-box" model and details related to the CIREN crash sample. Finally, a novel quantitative computational model of injury severity was used to assess how factors of the crash event influence the loading mechanisms of severe distal tibia fractures.
Assessing Career Trajectories: Comparative Analysis of Career Outcomes of International and US-born College Graduates in the US Labor Market
Jan, Baby Faika Tahir (Virginia Tech, 2026-08-24)
Globalization has produced disjunctive flows and new global cultural configurations, heralding an era of unprecedented connectivity and diversity. International college graduates have become an essential element of service- and technology-oriented economies such as the US, covering labor gaps and keeping wages stabilized. This research analyzes the confluence of identities among foreign-born college graduates in the context of their career outcomes in comparison to US-born college graduates, accounting for their intersectional identity characteristics. Using data from the 2021 National Survey of College Graduates (NSCG), this study aims to identify intersectional and labor market factors that influence career outcomes such as salary, relevance of the job to one's highest degree (hereafter termed major-job match), and job satisfaction among international and US-born graduates in the US labor market. By addressing this research questions and hypotheses, this dissertation aims to better understand international college graduate workers' career trajectories in the US labor market and foster inclusive and equitable career opportunities for an increasingly diverse cohort of graduates in the U.S. labor market.
Data-Driven Prokaryotic Genome Identification: LIN Assignment, Taxonomy Correspondence, and Deployment
Mazloom, Reza (Virginia Tech, 2026-08-21)
This dissertation addresses a central challenge in microbial genomics: how to organize and identify rapidly growing genome collections in a way that is both computationally scalable and biologically interpretable. It presents a data-driven framework centered on Life Identification Numbers (LINs), where each genome receives a hierarchical, threshold-based label, and shared prefixes encode relatedness at multiple resolutions. The work develops LINflow 2.0, a modular assignment system that combines high-throughput sketch-based filtering with average nucleotide identity refinement to support efficient, incremental LIN assignment at database scale. LIN-derived clusters are then compared against NCBI and GTDB taxonomies, showing strong correspondence at many ranks, while also revealing lineage-specific discordances that can guide recircumscription. The framework is operationalized in genomeRxiv, a web platform that supports rapid genome or sketch submission, query assignment, taxonomy circumscriptions, and search across related genomes. Together, these theoretical, algorithmic, and platform contributions provide a unified approach that bridges traditional taxonomy and strain-level typing, enabling faster and more reproducible genome identification for microbial research, surveillance, and outbreak response.
Multibody Dynamics of Aeromechanical Systems
Zakaria, Mohamed Ahmed Yehia (Virginia Tech, 2026-08-21)
Gliding flight is frequently treated as a reduced problem in rigid body aerodynamics, but many natural and engineered gliders undergo internal shape change, appendage motion, and coupled body deformation that substantially alter their dynamics. These features make gliding a problem that lies at the intersection of biomechanics, aerodynamics, multibody dynamics, nonlinear dynamical systems, and control. Although important progress has been made within each of these areas, the development of a unified framework that connects them remains limited. This dissertation aims to contribute to such a framework by examining how internal motion and geometry change influence the dynamics, stability, and control of gliding systems across both biological and engineering contexts. To this end, the dissertation studies several related manifestations of coupled glide dynamics. It first investigates the role of tail motion in gliding lizards, showing that the tail contributes to longitudinal response through both inertial and aerodynamic mechanisms and thereby influences pitch behavior and stability. It then considers lateral aerodynamic effects through the analysis of side force in textit{Draco} glides, demonstrating that glide motion is inherently three dimensional and that its aerodynamic response depends on both configuration and time varying kinematics. The work further develops a three dimensional dynamical systems framework for passive gliding, in which terminal velocity manifolds and separatrix structures organize glide trajectories and provide a geometric interpretation of distinct descent behaviors. Overall, the dissertation shows that gliding systems are most naturally understood as coupled multibody dynamical systems in which internal geometry and appendage motion play a fundamental role in force production, stability, and control. In doing so, it contributes a unified perspective that connects biological gliding, dynamical systems analysis, and morphing aircraft design.
From Diary to Discussion: The PCAR Framework for Scaffolding Empathy-Oriented Learning in HCI Education
Fan, Jixiang (Virginia Tech, 2026-08-19)
Computing education has long emphasized functional implementation, technical correctness, and system building. As computing systems become increasingly embedded in real-world social contexts and everyday life, however, helping students understand user experience, contextual differences, needs and constraints, and the human impact of technical decisions has become an increasingly important educational goal. Yet for many undergraduate computing students, it remains challenging to shift from technical implementation to user understanding in human-computer interaction and to translate user needs into concrete system design decisions. Design empathy is an important capability for supporting this shift. Here, empathy does not simply refer to emotional resonance, but to the ability to understand users' situations, goals, constraints, and value judgments, and to bring that understanding into design thinking. This dissertation introduces diary study as a pedagogical approach in HCI and computing education, with the goal of helping students gradually develop user-oriented understanding and design judgment through sustained recording, analysis, and reflection on technology-use experiences. Through an exploratory comparison of conventional HCI learning activities, represented by observation and affinity diagramming, with diary recording alone, this dissertation finds that conventional HCI activities do not necessarily lead to stable gains in students' empathy, while sustained diary recording shows potential as an instructional scaffold. Building on this exploration and on established diary study practices in HCI, this dissertation translates the diary study research method into the Plan, Collect, Analyze, Reflect (PCAR) diary study framework. This framework supports students in turning real-world experiences into learning materials that can be recorded, analyzed, and discussed. In the Reflect stage of PCAR, this dissertation further designs and implements a reflection activity centered on focus group discussion, forming a diary-to-discussion learning process that moves from individual experience recording to collective reflection. Diary recording provides students with sustained, concrete, and contextualized materials from their own use experiences, allowing them to identify problems, emotional responses, and potential user needs through real interactions. Focus group discussion then gives students opportunities to encounter others' experiences, compare differences, revise judgments, and reconsider user contexts. To evaluate this learning process, the study uses the EMPA-D scale to measure changes in students' empathy at three stages: before the learning activities, after diary recording, and after focus group discussion. It also analyzes students' reflective feedback through qualitative coding. The results show that students' empathy-related abilities improved significantly after diary recording and continued to develop after focus group discussion. The qualitative findings further show that students not only reinforced and refined their existing user awareness but also deepened their understanding of users, contexts, and design tradeoffs while beginning to reexamine their own assumptions, judgments, and cognitive blind spots. Overall, this dissertation proposes and evaluates a PCAR-based diary study pathway for supporting empathy-oriented learning in HCI education, and offers a structured and practical pedagogical approach for human-centered learning in broader computing education.
Pulse-Driven Microfluidic Pumps for Transdermal Drug Delivery and Material Investigation of PDMS and 3D Printing Resins
Zhang, Shuyu (Virginia Tech, 2026-08-18)
Nowadays, pharmaceuticals such as insulin, vaccines, chemotherapeutics, and analgesics still need to be delivered across the skin using relatively invasive subcutaneous cannulas coupled with either powered pumps or syringes. Although these techniques have been shown to be promising over the years, they have significant limitations, including but not limited to injection site pain and discomfort, infections, interference with daily activities, cost, and embarrassment. Considering this, we have developed a novel technology of wrist-worn, pulse-driven microfluidic pumps. We fabricated our devices with polydimethylsiloxane (PDMS), refined our device prototype in 12 design generations, and evaluated the performance of these pumps. We examined the flow rate generated by these devices using a pressurized air pulse simulator and on the wrists of 116 healthy human subjects. Linear mixed models that correlate the pulse-driven flow rate with device and user parameters were constructed, suggesting that the flow rate generated was positively correlated with the width and membrane area of the main flow channel and enhanced by wrist motion. In addition, both wicking with polyester wipes and transdermal delivery with porcine skin significantly increased the flow rate produced by our devices. We developed and performed three related studies to better understand how material properties potentially impact the performance of our devices. Various designs of 3D-printed hollow microneedle arrays were fabricated and investigated for transdermal penetration and mechanical strength. We found that syringe-shaped microneedle arrays with Formlabs Surgical Guide resin can effectively penetrate porcine skin without a risk of microneedle fracture. In addition, we filled a gap in research by characterizing the natural aging of PDMS at five base-to-curing agent mixing ratios and six mild storage environments at room temperature. PDMS at all mixing ratios was found to slightly increase in surface hydrophobicity, and PDMS in some mixing ratios and storage environments exhibited drastically increased stiffness and brittleness. We found that storage underwater can help preserve hydrophilicity and mechanical strength of PDMS. We also tested the effect of post-print treatment conditions of stereolithographic (SLA) 3D printing resins on the curing of their PDMS replica, which suggested that heat treatment followed by a post-washing step of negative master molds can enhance the channel geometry fidelity, surface smoothness, and mechanical strength of cast-molded PDMS. Throughout this project, we have collected preliminary data to confirm the feasibility of our devices as a platform technology for drug delivery. Future work includes developing our technology for specific drug delivery applications and for cardiac waveform detection.
GeoAI framework for crop nutrient and above ground biomass estimation using UAV multispectral remote sensing and machine learning across diverse Virginia environments
Kumari, Sheetal (Virginia Tech, 2026-08-18)
Accurate, timely, and spatially explicit estimation of crop biomass and nutrient status is essential for precision nutrient management in winter wheat (Triticum aestivum L.). Conventional assessment relies on destructive tissue sampling and laboratory analysis, which is labor-intensive, costly, and unable to capture field-scale spatial variability within the time frame required for in-season decisions. This dissertation addresses these limitations by developing a consistent unmanned aerial vehicle (UAV) multispectral remote sensing and machine learning (ML) framework for estimating aboveground biomass (AGB), grain yield, and the uptake of nitrogen (N) and phosphorus (P). The framework integrates spectral vegetation indices (VIs), gray-level co-occurrence matrix (GLCM) texture, normalized difference texture indices (NDTI), and ancillary variables within multiple linear regression (MLR) and random forest (RF) models. It was evaluated across 176 plots at three contrasting Virginia field sites during the 2024-2025 growing season. The research began with a systematic map of remote sensing algorithms for N and P estimation. This synthesis revealed a pronounced technological divergence: N estimation has matured into a reliable tool, whereas P estimation remains a substantial and largely unresolved challenge, attempted in only a small minority of studies. These findings defined the gaps that the empirical chapters were designed to address. The first empirical study estimated AGB and grain yield. Random forest combined with NDTI cross-band texture achieved its highest accuracy at the tillering stage (R2= 0.84; RPD=2.57), establishing tillering as the optimal imaging window, with accuracy declining through later stages as the canopy saturated and senesced. Direct yield prediction proved unreliable due to lodging-induced harvest losses but applying site-specific harvest index (HI) values to AGB maps produced credible spatially continuous yield estimates. The second empirical study estimated P uptake. Estimation was most reliable at tillering, where texture and cross-band features captured P-driven structural variation that spectral indices alone could not represent (RF R2=0.86; RPD=2.77), while remaining challenging during canopy transition at booting. The resulting P uptake maps resolved within-field variability and differences attributable to contrasting fertilization histories. The third empirical study estimated plant N accumulation. Reliable predictions were obtained at tillering (RF R2=0.84; RPD=2.67) and at a useful level at booting, whereas at heading poor prediction was likely due to N values convergence at canopy closure. Notably, ground-measured ancillary variables, including plant height, chlorophyll readings, and soil chemistry, added little once imagery features were available, supporting a workflow based on UAV imagery alone. These studies demonstrate that a single, consistent UAV multispectral framework can estimate AGB, nutrient status, and yield in winter wheat. The estimation accuracy is governed in a systematic and interpretable manner by growth stage, feature type, and the nutrient concerned. Tillering was consistently the most informative stage, texture features and RF contributed most strongly to the early season, and N proved far more tractable than P. By clarifying when and why UAV-based estimation succeeds, and by establishing a largely imagery-based workflow validated across diverse environments, this research provides both new scientific understanding and practical tools for precision nutrient management in Mid-Atlantic winter wheat production.
Emission of Microplastics via Sea Spray Aerosol: A Generalizable Enrichment Framework for Marine Constituents
Pokhrel, Nishan (Virginia Tech, 2026-08-18)
Micro- and nanoplastics (MNPs) are emerging atmospheric pollutants, yet the processes controlling their transfer from the ocean to the atmosphere remain poorly constrained. Sea spray aerosol (SSA), produced when bubbles generated by breaking waves burst at the ocean surface, provides a plausible pathway for transferring MNPs and other marine constituents from seawater to air. Given their potential for inhalation exposure, long-range atmospheric transport, and deposition into remote ecosystems, constraining oceanic MNP emissions is necessary for evaluating their environmental fate, exposure pathways, and potential public-health risks. However, current understanding of this transfer is limited by uncertainty in how MNP enrichment in emitted droplets depends on particle properties, droplet production pathways, and environmental parameters. This dissertation addresses these gaps by combining controlled laboratory experiments with a pathway-resolved framework for representing oceanic emissions of MNPs and other preferentially enriched marine aerosol constituents in atmospheric models. First, MNP aerosolization was quantified in a Marine Aerosol Reference Tank (MART) using MNPs of different sizes, densities, and concentrations in synthetic seawater. These experiments showed that MNPs with diameters up to $10 mu m$ can be emitted via bubble bursting, that aerosolization increases with particle concentration in water and decreases with increasing particle size, and that particle density and water-column distribution influence transfer efficiency. The resulting experimentally based parameterization provided an initial constraint on global oceanic MNP emissions. Second, this dissertation addressed the need for droplet pathway separation in sea spray source functions, which are mathematical representations used in atmospheric models to estimate size-resolved SSA emission. When bubbles burst at the water surface, they can produce film drops through fragmentation of the bubble cap and jet drops through collapse of the bubble cavity. Because film and jet drops differ in their production mechanisms, size ranges, and enrichment behavior, treating SSA production as a single bulk size-resolved flux can be insufficient for marine constituents whose transfer depends on droplet production pathway. A Spray Aerosol Pathway Tank (SAPT) was therefore used to generate distinct bubble populations associated with film- and jet-drop-dominated aerosol production and to develop a film- and jet-drop-resolved sea spray source function. This source function represents SSA production as separate film and jet components and enables pathway-specific enrichment factors to be coupled with pathway- and size-resolved aerosol fluxes. The resulting pathway-resolved source function curve was broadly consistent with the range of widely used sea spray source functions, while additionally resolving the film and jet drop contributions needed to represent pathway-dependent enrichment. Demonstration calculations for the emission flux of dissolved organic carbon, calcium, saccharides, bacteria, MNPs, and per- and polyfluoroalkyl substances showed that pathway resolution can substantially alter the magnitude and size distribution of predicted marine constituent emissions. The dissertation then examined how MNP surface and environmental properties influence pathway-resolved transfer. Experiments with spherical $1 mu m$ surface-modified polystyrene particles showed that hydrophobic particles were enriched in jet drops by approximately one order of magnitude more than hydrophilic particles, whereas no clear wettability dependence was observed for film drops, suggesting that MNP particles may have complex effects on bubble film stability, bursting, and enrichment dynamics. Finally, experiments with environmentally representative MNPs showed that mechanically fragmented polylactic acid particles were not aerosolized as a random subset of the water population; instead, film and jet drops preferentially transferred smaller and more compact fragments. Experiments with pristine, mechanically abraded, and biofouled $8 mu m$ polystyrene particles further showed that enrichment increased after mechanical abrasion and increased further after biofouling. Overall, this dissertation demonstrates that oceanic bubble bursting can transfer MNPs from seawater to air and that this transfer depends on particle size, concentration, density, wettability, morphology, biofouling state, and droplet production pathway. By linking laboratory measurements of pathway-separated MNP aerosolization with a film- and jet-drop-resolved SSA source function, this work provides a physically informed framework for improving oceanic MNP emission estimates and for representing other preferentially enriched marine aerosol constituents in atmospheric models.

Browse

Recent Submissions