MSU Libraries Digital Repository (Michigan State University)
Not a member yet
106711 research outputs found
Sort by
DIAGNOSTICS, GENOMICS, AND MANAGEMENT OF DIAPORTHE HUMULICOLA, THE CAUSAL AGENT OF HALO BLIGHT OF HOPS
No full textThesis (Ph.D.)--Michigan State University. Plant Pathology - Doctor of Philosophy, 2025This dissertation explores various aspects of the pathogen D. humulicola. The research tested multiple fungicide products to determine effective management strategies for the disease. Additionally, a qPCR tool was developed for the detection of D. humulicola in different plant materials. Lastly, a genome was produced, and a population genetics study was conducted using D. humulciola isolates from the eastern United States and Canada. The first research objective of this dissertation was to determine if fungicides could be used for the treatment of halo blight. For three years field trials were conducted using 10 fungicides registered for use hop to determine their efficacy against halo blight. To validate field results, the EC50 value was determined for some of the active ingredient including flutriafol, tebuconazole + fluopyram, cyflufenamid, and trifloxystrobin + salicylhydroxamic acid (SHAM). A discriminatory dose was used to test the sensitivity of 206 D. humulicola isolates in a poison agar assay. Results showed that tebuconazole + fluopyram decreased the incidence and severity of halo blight in the field. Also, this product had EC50 values of 2.26 x 10-1 ppm and significantly reduced the growth of most of the isolates tested. Trifloxystrobin + SHAM decreased the presence of halo blight in the field, but some isolates were not inhibited by the discriminatory dose concentration. The second research objective was to develop a diagnostic tool for D. humulicola. A quantitative polymerase chain reaction (qPCR) assay based on the translation elongation factor 1-alpha gene was developed. We assessed this assay for direct detection of D. humulicola in plant tissue and investigated aspects of the disease cycle through three distinct experiments: 1) detection of D. humulicola in hop rhizomes to determine the colonization range of the pathogen, 2) determining how quickly can D. humulicola be detected in hop leaves post inoculation, and 3) monitoring the presence of D. humulicola in cones in a hop yard and comparing isolation methods and the assay. The limit of detection for the assay was 100 fg of DNA. The assay showed no cross-reactivity with other hop pathogens or endophytes, nor with other Diaporthe species tested. Detection of D. humulicola occurred one day after inoculation. The assay detected D. humulicola in both apparently healthy and diseased rhizome tissue. The assay successfully detected the pathogen in individual hop cones and inflorescences throughout the season, surpassing the culture-based method in positive identification rates. The third objective was to produce and annotate a genome for D. humulicola. The draft genome for D. humulicola, was assembled with both long and short read sequencing. The draft genome consists of 49.82 MB assembled into 180 contigs, with a GC content of 51.2%. The genome was annotated, and 11,773 genes were predicted, including 2,752 genes with common names. Finally, the genome was used to conduct a population genetics experiment using 64 different D. humulicola isolates from Michigan, New York, Minnesota, and Canada. Single nucleotide polymorphisms (SNPs) were discovered and filtered using GATK. Population Structure was determined using STRUCTURE v.2.3.4 and using the Evanno method the populations cluster into 4 or 6 different populations. Using Fst Minnesota isolates appear to have high levels of population differentiation when compared to the different populations. Mating types were determined for each isolate where Mat-1-2-1 were the larger part of the population with 59 percent of the isolates having this locus. This dissertation also includes an extension fact sheet for halo blight, and disease notes for different occurrences of halo blight in the United States and Canada.Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).Includes bibliographical references
HOLDING ONTO HERITAGE : THE PRACTICES, BARRIERS, AND BELIEFS OF ASIAN AMERICAN HERITAGE LANGUAGE LEARNERS
Get PDFThesis (M.A.)--Michigan State University. Teaching English to Speakers of Other Languages - Master of Arts, 2025Many Heritage Language Learners (HLLs) struggle to balance speaking their family\u2019s native language and English on a regular basis. Particularly in households of Asian descent, the struggle to speak fluently at home can cause HLLs heightened stress. This qualitative study, grounded in Asian Critical Theory (AsianCrit), explored the experiences of five Asian American HLLs: four Vietnamese speakers and one Mandarin speaker. Using semi-structured interviews and reflective journals, the study examined the challenges they face in maintaining their heritage language and the strategies they used to sustain proficiency. Key findings revealed that HLLs relied on family-based reinforcement, social engagement, digital learning tools, and community involvement to maintain their language. However, they also struggled with limited opportunities for practice, societal pressures, generational communication gaps, and difficulties with literacy, leading to language loss and frustration. The study called for greater community support, institutional recognition, and inclusive pedagogical strategies to ensure that heritage languages are preserved and valued in multilingual societies.Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).Includes bibliographical references
SECURITY ENHANCEMENTS FOR LLM-BASED CHATBOTS
Get PDFThesis (Ph.D.)--Michigan State University. Computer Science - Doctor of Philosophy, 2025Large Language Model (LLM)-based chatbot systems have transformed various aspects of everyday life, including smart home devices, recommendation systems, and search engines. As the capacity and integration of LLMs expand, these systems facilitate daily communications, assist with specialized tasks such as medical documentation, and even operate real-world transactions like event bookings or financial management. However, the probabilistic nature of LLMs and their reliance on human-curated training data introduce substantial security and privacy concerns. This dissertation identifies vulnerabilities in LLM-based chatbot systems and proposes practical and effective attack and defense strategies. The research spans both the training phase and the inference phase of chatbot system development and deployment, guided by the dual principles of system-level practicality and model-level vulnerability.First, this dissertation investigates a vulnerability in which an adversary can corrupt training data and inject backdoors into end-to-end LLM-based chatbot systems. Prior backdoor attacks have targeted single-turn tasks or relied on access to explicit data labels. This dissertation introduces a novel, label-free multi-turn backdoor attack, embedding subtle triggers, such as natural interjections, across several conversational turns in the fine-tuning data. Experimental evaluations demonstrate that poisoning less than 2\% of the training data enables adversaries to control LLM outputs when specific triggers are present, while maintaining stealthiness. Second, to protect user data privacy during model training, a dynamic federated learning framework is proposed. Existing federated learning client selection methods often suffer from low communication efficiency, resulting in suboptimal training time and wasted resources. This dissertation introduces a new framework that incorporates bandwidth prediction and adaptive client scheduling into federated learning for LLM training. By leveraging a long-term observation window and predictive modeling of client network conditions, the framework effectively selects the most reliable clients, ensuring faster training convergence and improved time-to-accuracy without compromising user privacy. The proposed method outperforms prior approaches under fluctuating real-world network scenarios. Third, this dissertation investigates the vulnerability of chatbot systems that generate toxic responses in user interactions. Existing evaluation efforts primarily focus on single-turn prompts, overlooking the dynamic escalation of harmful content in multi-turn dialogues. We introduce an LLM-based red teaming tool that automatically and agnostically engages in multi-turn conversations with the target model to elicit harmful outputs. We find that seemingly non-toxic individual sentences can trigger toxic behavior in conversation, and these are often classified as safe by existing tools. Fourth, this dissertation investigates practical defenses that can be implemented when only API access is available. We propose a low-cost plugin, the Moving Target Defense, to enhance LLM-based chatbot systems against jailbreak attacks. This approach dynamically adjusts decoding hyperparameters at inference time, which are settings that control randomness in next-word prediction. By modifying these parameters, the system can reject jailbreak attacks while maintaining utility and responsiveness. Importantly, this defense requires no access to the underlying model weights or retraining, and can be deployed as a post-processing layer for existing LLM APIs. Experimental results show substantial improvements in defense effectiveness compared to static black-box defense strategies.Overall, these research threads reveal a broad and urgent attack surface for LLM-based chatbot systems, ranging from subtle poisoning of training data to inference-time adversarial exploitation. Through the development and evaluation of both attack and defense techniques, this dissertation not only uncovers current vulnerabilities, but also advances mitigation technologies suitable for real-world deployment.Finally, the dissertation concludes with a discussion of future research directions, emphasizing the ongoing evolution of LLM architectures, the challenging arms race between attacks and defenses, and the necessity for systematic evaluation frameworks. As LLM-based chatbots continue to improve across critical sectors, the foundational discoveries and proposed mechanisms in this work aim to support a more secure, AI-enabled future.Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).Includes bibliographical references
A SCOPING REVIEW OF RISK FACTORS FOR PREGNANCY-ASSOCIATED THYROID CANCER
Get PDFThesis (M.S.)--Michigan State University. Epidemiology - Master of Science, 2025Background: Pregnancy-associated cancers affect 1 in 1000 women globally with breast cancer, melanoma, and thyroid cancer being the most diagnosed. While extensive research exists on pregnancy-associated breast cancer, studies on pregnancy-associated thyroid cancer (PATC) remain limited. This scoping review aimed to summarize existing evidence on PATC risk factors, and whether these risk factors differed from thyroid cancer (TC) not diagnosed during pregnancy, as well as identify gaps in current knowledge to guide future research. Peer-reviewed studies published in English-language journals investigating exposures in females aged 15 to 50 years old with thyroid cancer were eligible for inclusion in the review.Methods: A PubMed search was not confined by a specific publication date range. The 185 identified articles published between 1987 and 2024 were imported into Covidence for review and extraction. The \u2018Title and Abstract\u2019 review, \u2018Full Text\u2019 review, and data extraction were completed in duplicate by three reviewers. Conflicts were resolved through discussion until a consensus was reached. Thirteen studies were included in the scoping review eligible for extraction. Post-extraction consensus was reached by one reviewer per the rules of Covidence Extraction Tool 1, which allowed for pre-extraction control and update of the data template. Qualitative analyses were performed. Synthesis followed.Results: Three studies showed significant odds for developing PATC with the presence of palpable goiter or thyroid nodule (OR=20.5), history of gestational diabetes (GDM) (OR=1.24-4.3), and positive blood tests for thyroid peroxidase antibody (TPO-Ab) (OR=3.32) and thyroid globulin antibody (Tg-Ab) (OR=2.03). Only four variables were evaluated in relation to both PATC and TC; parity, history of infertility, history of miscarriage or abortion, and oral contraceptive (OC) use. None were associated with PATC; they were mixed factors with TC. For TC, parities had an increasing significant trend (OR=1.19-3.0), history of infertility were nonsignificant (OR=1.6), history of miscarriage or abortion were nonsignificant (OR=1.1-2.7), and OC use had a mixed protective (OR=0.74) or risk (OR=1.72-3.8). Conclusions: Although, the risk factors for PATC and TC are similar according to the American Thyroid Association (ATA), findings from this review suggest palpable goiter/thyroid nodule, GDM, and autoimmunity markers as an additional PATC risk factor.Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).Includes bibliographical references
INVESTIGATION OF COMBINED ORAL CONTRACEPTION AND BACTERIAL VAGINOSIS IN US PREMENOPAUSAL INDIVIDUALS
Get PDFThesis (M.S.)--Michigan State University. Epidemiology - Master of Science, 2025ABSTRACTThe most common cause of vaginal discharge is bacterial vaginosis (BV). BV, a dysbiosis of the vaginal microbiome, is inhibited by lactic acid, and lactic acid production is supported by estrogen. Use of estrogen-containing combined oral contraception (COC) has been associated with decreased BV prevalence across several studies. However, most studies relied on self-report to ascertain COC use. We conducted a cross-sectional study of COC use and BV using data from the National Health and Nutrition Examination Survey (NHANES) 2001-2004 cycles; data on COC use was collected from two sources. During the household interview, participants were asked about prescription medication use in the past month. During the NHANES exam, in a private interview, participants were asked about the current use of birth control pills. Vaginal fluid was collected, Gram stained, and scored using the Nugent criteria for BV (scores of 0-3, 4-6, and 7-10 corresponded to negative, intermediate, and positive BV). Among non-pregnant, premenopausal individuals ages 20-49 (unweighted n=1460), we conducted multinomial logistic regression to estimate the adjusted odds ratios (aOR) and 95% confidence intervals (CI), accounting for the complex survey sampling design. The weighted prevalence of prescription medication-derived current COC use was 7%. Current COC use was associated with 67% lower odds of intermediate Nugent-score BV (aOR 0.33; 95%CI 0.18-0.63) and 83% lower odds of positive Nugent-score BV (aOR 0.17; 95%CI 0.07-0.42) compared to non-use. In contrast, the weighted prevalence of self-reported current OC use was 18%. For self-reported current OC use, we observed lower odds of intermediate Nugent-score BV (aOR 0.50, 95%CI 0.34-0.75) and positive Nugent-score BV (aOR 0.48; 95%CI 0.30-0.79). COC use based on prescription information yielded a stronger association than that based on self-report. Further research investigating the pharmaceutical aspects of COC use on BV prevalence is warranted.Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).Includes bibliographical references
Statistical Learning-Based Adaptive Attacks Towards Audio Watermarking
Get PDFThesis (M.S.)--Michigan State University. Computer Science - Master of Science, 2025The abuse of original audios has attracted widespread attention in the society. Audio watermarking, which embeds imperceptible signals into audio content, has been proposed as an effective way to assert user copyright of audios. Although recent deep learning-based audio watermarking methods have enhanced robustness and capacity compared to traditional approaches, they are vulnerable to adversarial attacks. Our findings reveal that the message probabilities output by the watermark decoder follow a normal distribution for both clean and watermarked audio. This observation can be leveraged to detect existing audio watermark attacks. In this thesis, we introduce AWM, an adaptive audio watermark attack method designed to bypass existing detection strategies. The attack has three different types: watermark replacement, watermark creation, and watermark removal. AWM employs a two-step optimization process: the first step ensures the success of the watermark attack and bypasses the detection by optimizing message probabilities within an estimated normal range, while the second step focuses on enhancing audio quality while maintaining a successful attack. The proposed attack iteratively estimates the parameters of the normal distribution using a small set of feature-similar audio samples based on the target audio and applies adaptive optimization to adjust the decoded message probabilities toward the estimated normal range. We evaluate AWM on two watermarking methods across three diverse voice datasets and compare the results with existing audio watermark attack techniques. Our experiments demonstrate that the proposed attack achieves a high attack success rate while effectively bypassing detection, with detection success rates remaining under 10% for watermark replacement and watermark creation, and at 0% for watermark removal. Additionally, AWM exhibits high robustness against various no-box perturbations, including low-pass filtering, amplitude scaling, and compression, while maintaining high perceptual audio quality. Our experiments highlight a significant security gap in current watermark defenses and show that statistical assumptions about the decoder output can be exploited by attackers. These findings also provide a foundation for future research in audio watermark attack detection and the development of more advanced attacks.Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).Includes bibliographical references
THE EFFECTS OF VIBRATIONAL THERAPY ON ANXIETY AND SLEEP QUALITY IN YOUTH WITH ADHD : A DOUBLE-BLIND, RANDOMIZED SHAM-CONTROLLED TRIAL
Get PDFThesis (Ph.D.)--Michigan State University. Kinesiology - Doctor of Philosophy, 2025ABSTRACTAttention Deficit Hyperactivity Disorder (ADHD) is one of the most prevalent neurodevelopmental disorders and is associated with significant impairments in cognitive functioning. Moreover, the disorder is associated with increased anxiety and poor quality sleep, behaviors which may exacerbate and/or mimic ADHD symptomatology. As the autonomic nervous system (ANS) plays a critical role in modulating physiological levels of arousal, therapeutic strategies which target autonomic balance, such as vibrational therapy, may offer substantial benefits for anxiety and sleep. The Apollo system, a wearable vibrational therapy device designed for use on the wrist or ankle, has been found to promote greater autonomic balance by enhancing parasympathetic tone. However, its effects on sleep quality and anxiety levels in children with ADHD remain unclear. Objective: Therefore, the current study employed a double-blind, randomized sham-controlled design to evaluate the impact of the Apollo system on anxiety levels and sleep quality in children and adolescents (aged 8 to 17 years) with ADHD. Method: Participants were randomly assigned to either the active Apollo device, which delivers vibration to promote autonomic balance, or a sham device over a period of 8 weeks. Anxiety and sleep were assessed prior to and following this intervention period, using the Pittsburgh Sleep Quality Index (PSQI) and the trait subset of the State-Trait Anxiety Inventory (STAI). Conclusion: The findings indicate that utilizing a wrist/ankle-based vibrational therapy device 4 days per week for 2 hours a day over an 8-week period may be insufficient to meaningfully improve sleep outcomes and anxiety in children and adolescents with ADHD.Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).Includes bibliographical references
White Educational Resource Hoarding in a Small New England Town
Get PDFThesis (Ph.D.)--Michigan State University. Curriculum, Instruction, and Teacher Education - Doctor of Philosophy, 2025Dominant understandings of \u201cgood\u201d schools are strongly associated with the presence of white students. The purpose of this dissertation is to denaturalize the association of goodness and whiteness in U.S. schools, in order to illuminate how this association, which is deeply embedded in understandings of meritocracy and deservedness, perpetuates injustice in and through U.S. schools. For this study, I chose the small, Massachusetts town of Cobblerton (pseudonym), because of its overwhelming whiteness, reputation for excellent schools, proximity to majority Black and Brown Boston Public Schools, and because I grew up in Cobblerton and graduated from its high school. I employ a critical race methodology to build on and extend current research on school segregation and inequality by investigating the causes and consequences of what I call \u201cwhite educational resource hoarding.\u201d This study asks: How was Cobblerton\u2019s whiteness created and maintained historically? What are the impacts of white educational resource hoarding on Cobblerton and its schools? How do narratives of the \u201cgoodness\u201d of Cobblerton and Cobblerton Schools justify and sustain white educational resource hoarding?Through interviews, both with key town decisionmakers and those impacted by their decisions, as well as historical document analysis, and observations of school committee and strategic planning meetings, I interrogate the historical and contemporary processes by which this small Massachusetts school district created and sustained itself as an overwhelmingly white space and at that same time, developed a reputation for school excellence. I also employ autoethnographic methods to analyze my own experiences as a student in Cobblerton and, later, as a teacher of mostly Black, Brown, and Indigenous students, in the context of white educational resource hoarding. I find that white educational resource hoarding is the process by which purportedly "good" educational resources, experiences, and outcomes accumulate in educational spaces that are accessible to white students but exclusionary to Students of Color, thereby normalizing and reproducing the deeply held association between good schools and whiteness. While this process involves policies and practices directly related to schooling, it also includes dynamics of infrastructure development, zoning, and town planning. I find that policies in the early 1970s were explicitly created to protect the town\u2019s whiteness, but residents today are unaware of this, which makes the town\u2019s whiteness seem natural, normal, and innocent. As a result, the pedagogy, curriculum, and culture of the \u201cgood\u201d school district is white-normed, and Students of Color suffer from a lack of culturally responsive pedagogy, curricular erasure, and racist bullying, which the school is unprepared to respond to or prevent. Further, district leadership pursues white-normed understandings of goodness for their schools that contain no critique of systemic injustice, and so seek to manage rather than abolish inequality. This research into white educational resource hoarding has implications for educators, including teacher educators, school districts, and education researchers. For educators, this work intensifies the urgency of adopting culturally relevant pedagogy, especially the critical consciousness aspect, in order to disrupt the white ignorance that is otherwise reproduced through schools like Cobblerton\u2019s. School districts can support teacher development as critical educators. In so doing, they will be pushed to reconsider how to define \u201cgoodness\u201d within their districts. Finally, education researchers must untangle our understandings of good schools from whiteness, and critically ask and keep asking, what is a good school? What is a good education, under racial capitalism, during a climate crisis and a U.S.-sponsored genocide, on violently stolen land?Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).Includes bibliographical references
BRIDGING AUTISM EVALUATION GAPS : A PARENT TRAINING APPROACH
Get PDFThesis (Ph.D.)--Michigan State University. School Psychology - Doctor of Philosophy, 2025Caregivers of children awaiting an evaluation for autism undergo increased levels of stress (Mulligan et al., 2012; Zukerman et al., 2015) and their self-efficacy with parenting is frequently impacted (Murphy & Harrison, 2022; Tang et al., 2015). Most interventions to date have focused on child outcomes while waiting for an evaluation for autism (Zukerman et al., 2015). Very few studies have supported caregivers during the waiting list, while also measuring their outcomes (DesChamps, 2020). The present study examined the initial efficacy of C-HOPE as an interim service for caregivers of children waiting for an evaluation for autism spectrum disorder. A single group pre/post method was used to assess whether participation in C-HOPE reduced caregivers stress levels, increased caregiver self-efficacy, increased knowledge about behavior management and behavioral principles, and reduced child problem behaviors. A secondary aim was to measure the acceptability, feasibility, and appropriateness of the intervention. The results of this study provide preliminary but promising evidence supporting the efficacy and acceptability of the C-HOPE parent training program for caregivers of children awaiting an evaluation for autism. Statistically significant changes were observed across several outcome variables, including reductions in caregiver stress, increased caregiver satisfaction, increased knowledge about behavioral principles, and decreased child problem behavior intensity, with large effects sizes each. No significant effects were observed for caregiver self-efficacy. There was an increase in efficacy on average, with a positive effect. Effects of C-HOPE on efficacy should be examined in future studies. Study implications and future research directions are providedDescription based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).Includes bibliographical references
DECIPHERING DITERPENOID BIOSYNTHESIS AND GENOME ARCHITECTURE IN THE LAMIACEAE FAMILY : INSIGHTS FROM CALLICARPA AMERICANA AND TEUCRIUM CHAMAEDRYS
Get PDFThesis (Ph.D.)--Michigan State University. Genetics and Genome Sciences \u2013 Doctor of Philosophy, 2025The Lamiaceae is one of the largest plant families incorporating over 7,000 species. It is home to a variety of ornamental and medicinal plants, but perhaps the most well-known are the culinary herbs, such as basil, oregano, rosemary, and thyme. The often-fragrant leaves of these plants are in part thanks to their terpenoid content. Terpenes are a class of natural compounds that are highly prolific in plants, and specifically the Lamiaceae family. These specialized metabolites can serve a variety of functions in the plant, including communication and defense. Terpenoids have been co-opted for human applications, as many can be used as medicines, pesticides, flavors, and fragrances. Occasionally, specialized metabolism, or specifically terpenoid metabolism, can be physically clustered in the genome. Biosynthetic gene clusters (BGCs) are a set of genes in the genome that are biochemically associated, phylogenetically distinct, and are often transcriptionally linked. One piece of the puzzle making the discovery of BGCs more feasible is the increase in robustness and quality of genome sequencing today. The following dissertation integrates the ideas of terpenoid metabolism and BGCs in the Lamiaceae family. Specifically, I explore the genomes of several Lamiaceae species, including Callicarpa americana and Teucrium chamaedrys, in search of BGCs and to explore terpenoid metabolism. I discovered a Lamiaceae family-wide gene cluster in C. americana that appears to predate the family at around 65 million years old. Additionally, I sequence and analyze the genome of the medicinally relevant plant T. chamaedrys, and find not only the same large BGC, but also discover the most putative diTPSs identified in a single species to date. I also functionally characterize several diTPSs and associated Cytochrome P450s. My work presented here aims to lay another brick in the larger building of understanding plant genomes and elucidating plant specialized metabolism.Description based on online resource. Title from PDF t.p. (Michigan State University Fedora Repository, viewed ).Includes bibliographical references