Digital Commons Network^™

Privacy-Preserving Collaborative Population Stratification With Dynamic Algorithm And Hyperparameter Selection

Faculty, Staff and Student Publications

We present a privacy-preserving selection layer for collaborative population stratification under 𝜖-local differential privacy (LDP). Rather than fixing a single pipeline (e.g., PCA+K-Means with preset 𝐾), our framework lets parties choose among three DP pipelines: PCA→Noise, Noise→PCA, and Noise-Only, according to their resources, and has an honest-but-curious server aggregate only DP shares to automatically select the clustering algorithm (K-Means, GMM, or Hierarchical) and 𝐾 that maximize internal metrics (Silhouette, Calinski–Harabasz, Davies–Bouldin). Because selection operates on DP data, it adds no further privacy loss. On openSNP (942 samples, 28,396 SNPs), the PCA-augmented pipelines yield higher utility …

On The Empirical Power Of Goodness-Of-Fit Tests In Watermark Detection, Weijie Su, Qi Long

Faculty, Staff and Student Publications

Large language models (LLMs) raise concerns about content authenticity and integrity because they can generate human-like text at scale. Text watermarks, which embed detectable statistical signals into generated text, offer a provable way to verify content origin. Many detection methods rely on pivotal statistics that are i.i.d. under human-written text, making goodness-of-fit (GoF) tests a natural tool for watermark detection. However, GoF tests remain largely underexplored in this setting. In this paper, we systematically evaluate eight GoF tests across three popular watermarking schemes, using three open-source LLMs, two datasets, various generation temperatures, and multiple post-editing methods. We find that general …

Privacy-Preserving Verification Of Ml Preprocessing Via Model Behavior Indicators, Erman Ayday

Faculty, Staff and Student Publications

We present a privacy-preserving framework to verify whether a declared data preprocessing pipeline was correctly applied before training a machine learning model on sensitive data. The verifier has only black-box query access to the model and combines three behavior indicators: shift in prediction accuracy, Kullback-Leibler (KL) divergence between output distributions, and explanation vectors from Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP). The method requires neither the original training records nor ground-truth labels. It supports two tasks: (i) a binary decision on correctness and (ii) a multi-class diagnosis identifying which step is missing. Experiments on three tabular datasets …

Safeguarding Privacy In Genome Research: A Comprehensive Framework For Authors

Faculty, Staff and Student Publications

As genomic research continues to advance, sharing of genomic data and research outcomes has become increasingly important for fostering collaboration and accelerating scientific discovery. However, such data sharing must be balanced with the need to protect the privacy of individuals whose genetic information is being utilized. This paper presents a bidirectional framework for evaluating privacy risks associated with data shared (both in terms of summary statistics and research datasets) in genomic research papers, particularly focusing on re-identification risks such as membership inference attacks (MIA). The framework consists of a structured workflow that begins with a questionnaire designed to capture researchers' …

Not Fully Synthetic: Llm-Based Hybrid Approaches Towards Privacy-Preserving Clinical Note Sharing

Faculty, Staff and Student Publications

The publication and sharing of clinical notes are crucial for healthcare research and innovation. However, privacy regulations such as HIPAA and GDPR pose significant challenges. While de-identification techniques aim to remove protected health information, they often fall short of achieving complete privacy protection. Similarly, the current state of synthetic clinical note generation can lack nuance and content coverage. To address these limitations, we propose an approach that combines de-identification, filtration, and synthetic clinical note generation. Variations of this approach currently retain 36%-61% of the original note's content and fill the remaining gaps using an LLM, ensuring high information coverage. We …

Enhancing Cross-Domain Generalizability In Social Determinants Of Health Extraction With Prompt-Tuning Large Language Models, Jiang Bian, Yonghui Wu

Faculty, Staff and Student Publications

The progress in natural language processing (NLP) using large language models (LLMs) has greatly improved patient information extraction from clinical narratives. However, most methods based on the fine-tuning strategy have limited transfer learning ability for cross-domain applications. This study proposed a novel approach that employs a soft prompt-based learning architecture, which introduces trainable prompts to guide LLMs toward desired outputs. We examined two types of LLM architectures, including encoder-only GatorTron and decoder-only GatorTronGPT, and evaluated their performance for the extraction of social determinants of health (SDoH) using a cross-institution dataset from the 2022 n2c2 challenge and a cross-disease dataset from …

Automatic Summarization Of Doctor-Patient Encounter Dialogues Using Large Language Model Through Prompt Tuning, Jiang Bian, Yonghui Wu

Faculty, Staff and Student Publications

Automatic text summarization (ATS) is an emerging technology to assist clinicians in providing continuous and coordinated care. This study presents an approach to summarize doctor-patient dialogues using generative large language models (LLMs). We developed prompt-tuning algorithms to instruct generative LLMs to summarize clinical text. We examined the prompt-tuning strategies, the size of soft prompts, and the few-short learning ability of GatorTronGPT, a generative clinical LLM developed using 277 billion clinical and general English words with up to 20 billion parameters. We compared GatorTronGPT with a previous solution based on fine-tuning of a widely used T5 model, using a clinical benchmark …

Exploring The Inequitable Impact Of Data Missingness On Fairness In Machine Learning

Faculty, Staff and Student Publications

Today, data-driven models and artificial intelligence / machine learning underlie decision making in almost all aspects of society. However, significant concerns have been raised over the fairness of such models. While various aspects of algorithmic fairness have been studied, the effect of missing data on fairness remains understudied. This is a significant problem since data in real-world settings is almost never complete, and may often suffer from systemic missingness. This article systematically evaluates how missing data, particularly when correlated with protected classes and outcome variables, affects the fairness of classifiers. Utilizing a comprehensive framework covering various missing data patterns, rates, …

Enhancing Medical Coding Efficiency Through Domain-Specific Fine-Tuned Large Language Models, Yan Zhuang

Faculty, Staff and Student Publications

Medical coding is essential for healthcare operations yet remains predominantly manual, error-prone (up to 20%), and costly (up to $18.2 billion annually). Although large language models (LLMs) have shown promise in natural language processing, their application to medical coding has produced limited accuracy. In this study, we evaluated whether fine-tuning LLMs with specialized ICD-10 knowledge can automate code generation across clinical documentation. We adopted a two-phase approach: initial fine-tuning using 74,260 ICD-10 code-description pairs, followed by enhanced training to address linguistic and lexical variations. Evaluations using a proprietary model (GPT-4o mini) on a cloud platform and an open-source model (Llama) …

Using Digitally Delivered Measurement-Based Care In Substance Use Disorder Treatment: Qualitative Analysis Of Patients’ Perspectives, Kevin A Hallgren

Faculty, Staff and Student Publications

Background: Measurement-based care (MBC) is a clinical practice where patients complete standardized outcome measures throughout treatment to monitor clinical progress and inform clinical decision-making. However, MBC is rarely adopted in routine substance use disorder (SUD) treatment. We developed a digital MBC system and pilot tested it in an outpatient adult community SUD treatment setting.

Objectives:The current study aims to characterize qualitative feedback from the pilot participants about their experiences using the MBC system in SUD treatment, focusing on perceived benefits, drawbacks, and suggestions for improvement.

Methods: Participants (N=30; n=11 female 37%) completed weekly MBC questionnaires via smartphone for 6 …

A Scoping Review Of Patient Safety Checklists In Pediatrics, Elsbeth Kalenderian, Donald B Rindal, Katie J Suda

Faculty, Staff and Student Publications

Objective: To examine the utilization and effectiveness of safety checklists in pediatric clinical care.

Methods: A comprehensive literature search was conducted using Medline to identify studies related to the development and/or implementation of patient safety checklists in pediatrics. All study designs were included for citations published through September 2023.

Results: Following abstract and full-text screening, 74 studies remained for data extraction and analysis. Pediatric surgery emerged as the main setting for checklists use (n = 35), followed by Intensive Care Units (n = 21), and Emergency Departments (n = 9). Of the 74 reviewed papers, 37 (50%) designed and developed …

Medication Prescribing Checklists And Their Impact On Patient Safety: A Scoping Review, Muhammad F Walji, Kristin N Ray, Donald B Rindal, Elsbeth Kalenderian, Katie J Suda

Faculty, Staff and Student Publications

Background: Medication prescribing checklists and the impact on patient safety remain unexplored.

Objectives: This scoping review aimed to identify key elements of these checklists and evaluate their impact on patient safety outcomes.

Methods: We searched PubMed and Embase for studies reporting checklists in medication prescribing practices and the impact on patient safety outcomes as of October 23, 2024. We applied Fisher's exact test to evaluate the association between checklist effectiveness (based on patient safety outcomes) with study design (randomized controlled trials [RCTs] vs non-RCTs) and intervention type (bundled intervention vs checklist-only intervention).

Results: After full-text review, 53 articles met inclusion …

Identifying Acute Myeloid Leukemia Subtypes Based On Pathway Enrichment, Xiaosheng Wang

Faculty, Staff and Student Publications

Acute myeloid leukemia (AML) is the most common type of acute leukemia in adults and the second most common in children. Despite the introduction of targeted therapies, AML survival rates have shown limited improvement, particularly among older patients. This study explored personalized treatment strategies for AML by proposing a novel subtyping method. Through unsupervised clustering based on the enrichment scores of 14 pathways related to metabolism, immunity, DNA repair, and oncogenic signaling, we identified three AML subtypes: DNA repair (DR), immune-enriched (ImE), and immune-deprived (ImD), consistent in four independent datasets. DR is marked by high expression of DNA repair and …

Depletion Of Adipose Stroma-Like Cancer-Associated Fibroblasts Potentiates Pancreatic Cancer Immunotherapy, Zhongming Zhao, Mikhail G Kolonin

Faculty, Staff and Student Publications

This study shows that populations of CAFs have distinct effects on pancreatic cancer progression and shows that depletion of CAFs expressing adipose markers potentiates tumor/metastasis suppression effects of immune checkpoint blockade.

Gene Expression Changes In Human Cerebral Arteries Following Hemoglobin Exposure: Implications For Vascular Responses In Sah, Devin W Mcbride, Kumar T Peeyush, Spiros L Blackburn

Faculty, Staff and Student Publications

Subarachnoid hemorrhage (SAH), characterized by the presence of hemoglobin (Hb) in the subarachnoid space, significantly impacts cerebral vessels, leading to various pathological outcomes. The toxicity of cell-free Hb released from erythrocytes and its metabolites after SAH causes vasoconstriction and neuronal damage, and correlates with delayed ischemic neurological deficits (DIND). While animal models have provided substantial and invaluable data in the research of aneurysmal SAH, the specific effects of subarachnoid blood on cerebral arteries remain greatly understudied. Here, we describe the changes in the genetic profile of human cerebral arteries exposed to free Hb for 48 h. We performed an ex …

Making Monsters: The Intersection Of Black And Queer Characters In Vampire Fiction, Jalissa Jones

Theses and Dissertations

For centuries we have looked at the vampire as a monster, one that skirts the line into acceptability not only because of the vampire's motivations, but more importantly because our depictions of vampires make them desirable to us. This thesis explores the vampire archive, and the question of how the monstrosity of the vampire interacts with race, sexuality, and gender to further uphold notions of white supremacy. This thesis will look at the classic vampire canon set forth by Bram Stoker’s Dracula and Carmilla by Sheridan Le Fanu and two contemporary black vampire narratives: The Gilda Stories by Jewelle Gomez …

Barriers To Mental Health Among International Students: Identifying Challenges And Developing Support Strategies For Depressive Symptoms, Esi Tb Afful

Theses and Dissertations

Objective: The purpose of this study was to examine barriers to mental health service utilization among international students, with a specific focus on challenges related to depressive symptoms and their coping mechanisms. The study also sought to identify factors that influence international students’ engagement with mental health resources.

Method: A total of 30 international students enrolled at a midwestern university participated in this mixed-methods study. Data were collected through an online survey, which included both closed- and open-ended questions. The quantitative portion of the survey assessed demographics, cultural and immigration challenges, mental health service utilization, and social support. Hierarchical multiple …

Enterprise Social Media And Employee Engagement: A Study Of Internal Public Relations, Daniel Sherrick Apambila

Theses and Dissertations

Workplace communication technology remains a tool for effective internal public relations practice in a postmodern business environment. Understanding how workplace communication tools function to enhance employees’ engagement, the role they play in forming employees’ perception of the company while fostering employees’ sense of belonging, when they are performing their organizational roles are critical to talent retention, sustained business operations, and maintaining organizational success in a competitive marketplace. However, research on the types of content posted on workplace communication platforms, and how the content shape intra-organizational members’ perception, while instilling a sense of belonging in their technical performance at the workplace …

Differential Effects Of Health Insurance Coverage On Health For Sexual Minorities, Ruby W. Morr

Theses and Dissertations

Much existing research has found a substantial gap in health between sexual minorities and non-minorities. This is usually attributed to discrimination or access to care. I seek to find the role that insurance plays, including minorities' choices of insurance coverage. If a gap does exist, insurance may have a stronger impact on sexual minorities. Also, if minorities are less likely to have insurance coverage, they may be susceptible to worse health. I confirm results from the literature that sexual minorities do have worse health. While health insurance accounts for some of the gap, it is only a small portion, though …

Lower Extremity Individual Muscle Forces In Females During Loaded Hex-Bar Squat Jumps, Jadon Konkel

Theses and Dissertations

The purpose of this study was to describe the individual muscle forces and contributions to extensor force occurring during the propulsive phase of unloaded jumps in females and explore how these are influenced by increasing loads using the hex bar modality. Ten, college-aged females (height: 1.66 ± 0.09 m; weight: 68.63 ± 9.66 kg) with weightlifting experience performed hexagonal bar squat jumps at bodyweight (BW), BW+20% and BW+40% of 1RM. Traditional motion capture techniques were employed to collect dominate lower limb kinematics and kinetics. Individual muscle forces were estimated utilizing Static Optimization in OpenSim using a Gait2392 musculoskeletal model. Muscle …

Enhancing Data Wrangling Efficiency Using Ai, Prajakta K. Pohare

Theses and Dissertations

Data wrangling is an important step in transforming raw data into the right format for the right analysis. Unfortunately, using the manual method can be drudgery and unreliable, not to mention error prone, especially with complex data sets. This study examines the role of AI in planning and organizing data wrangling tasks in the form of a Work Breakdown Structure (WBS) using an AI prioritizing tool. The research looks at how well and with how much breadth and flexibility the approach performs to deliver WBS prioritized tasks compared to a manually generated WBS prioritized tasks and deploy and test a …

Listening To The Body: Attunement And Advocacy In Women's Experiences Of Endometriosis, Rebekah E. Greenslaugh

Theses and Dissertations

Endometriosis is a chronic, inflammatory condition in which tissue similar to the endometrial lining grows outside the uterus, leading to severe pain, fatigue, and other debilitating symptoms. Despite its significant impact on daily life, many women with endometriosis report that their symptoms are dismissed in medical encounters, as their pain is minimized or normalized as an inherent aspect of womanhood. This medical dismissal, combined with a lack of definitive diagnostic tools and effective treatments, leads many to turn to online support communities to make sense of their symptoms, share experiences, and develop ways to cope with the condition. Drawing on …

Assessing Brazil’S Economic Trajectory Post Democracy: A Synthetic Control Approach To The 1985 Transition, Mohammad Saqlain Khurshid

Theses and Dissertations

Brazil’s democratic transition in 1985 marked a pivotal moment in its political and economic history. Following 21 years of military rule, the transition introduced sweeping political reforms amidst significant economic instability. This study examines the causal impact of democratization on Brazil’s economic trajectory using the Synthetic Control Method (SCM). By constructing a synthetic counterfactual, this research estimates the economic effects of democratization. The findings suggest a negative impact of democratization on the per capita GDP of Brazil, amounting to an average annual reduction of approximately $887 per year during 15 years following the democratization. However, this effect is not statistically …

A Bayesian Deep Segmentation Framework For Glioblastoma Tumor Segmentation Using Follow-Up Mris, Juan C Rodriguez Quintero, Octavio Arevalo, Kangyi Zhao, Jay-Jiguang Zhu, Roy F Riascos, Mahboubeh Madadi, Xiaoqian Jiang, Shayan Shams

Faculty, Staff and Student Publications

Background: Glioblastoma (GBM) is the most common malignant brain tumor with an abysmal prognosis. Since complete tumor cell removal is impossible due to the infiltrative nature of GBM, accurate measurement is paramount for GBM assessment. Preoperative magnetic resonance images (MRIs) are crucial for initial diagnosis and surgical planning, while follow-up MRIs are vital for evaluating treatment response. The structural changes in the brain caused by surgical and therapeutic measures create significant differences between preoperative and follow-up MRIs. In clinical research, advanced deep learning models trained on preoperative MRIs are often applied to assess follow-up scans, but their effectiveness in this …

Stimulating Focus And Calmness Through Soft Fabrication: A User-Centered Approach, Emmanuella A. Oku

Theses and Dissertations

This study explores the potential of active attention objects embedded with technologies such as LEDs to enhance focus and calmness in individuals with inattentive ADHD, particularly in classroom settings. Individuals with inattentive ADHD often struggle with irritability, stress, and panic attacks when exposed to external stimuli such as loud sounds, varying light intensities, strong smells, or physical sensations, which significantly hinders their ability to concentrate and remain composed. My literature review discovered a connection between an individual’s blinking pattern and their focus, and my work explores how to design active attention objects to engage with blinking as a focus tactic. …

“Don’T Panic, It’S Just Ai:” A Qualitative Study On Internal Communication Strategies For Managing Ai-Driven Change”, Edikan Nkanta

Theses and Dissertations

This qualitative study investigated how organizations utilize internal communication strategies to navigate AI-driven change effectively. Through semi-structured interviews with eleven communication professionals and change leaders across various industries, the research identified key communication approaches that facilitate successful AI implementation. Findings revealed three primary strategies: framing AI as a tool that enhances rather than replaces human capabilities, connecting AI initiatives to organizational purpose and business goals, and demonstrating practical use cases through concrete examples. The study also highlighted the importance of establishing clear guidelines for responsible AI use, making initial adoption optional rather than mandatory, and directly addressing job security concerns. …

On Higher Level Zhu Algebras Of N-Graded Vertex Algebras Associated With Simple Leibniz Algebras That Contain Sl2, Christian Soltermann

Theses and Dissertations

In this thesis, we study how the higher-level Zhu algebras of a vertex algebra reflect the structure of associated simple Leibniz algebras. In particular, we construct a vertex algebra from a vertex algebroid containing the simple Lie algebra sl₂ and analyze its higher level Zhu algebras. The irreducible modules of this vertex algebra were completely classified in [JY20b], but the structure of its indecomposable modules remains an open problem. Since modules for higher level Zhu algebras correspond to modules of vertex algebras, studying these algebras provides a method for understanding their broader representation theory.

Neural Network Algorithm And Analysis For Multi-Label Ecg Data Classification, Akhil Raghava Kalal

Theses and Dissertations

Electrocardiogram (ECG) analysis is a fundamental diagnostic tool in cardiology, providing critical insights into cardiac function that directly impact patient care decisions and treatment outcomes. As healthcare systems face increasing demands, automated ECG interpretation using artificial intelligence offers promising solutions to improve diagnostic accuracy, reduce physicians workload, and enhance early detection of life threatening conditions.

This thesis compares two advanced deep learning architectures, CNN-GRU and Wide and Deep Transformer, for multi-label classification of 12-lead ECG data. Using data from the PhysioNet/Computing in Cardiology Challenge 2020, I evaluated both architectures across 27 different cardiac abnormalities. Results demonstrated that CNN-GRU architecture consistently …

I Am Not A Dictator; I Am A Defender: The Rhetoric Of Coup D’État In Francophone Africa, Isaac Asare Kwakye

Theses and Dissertations

This study examines the rhetorical strategies used by coup leaders in Francophone Africa to authorize their actions and garner public support. Focusing on recent military takeovers in Mali, Burkina Faso, and Niger since 2020, I investigate how coup leaders construct narratives of legitimacy within the context of contemporary geopolitical dynamics. I utilize Critical Discourse Analysis (CDA) to analyze speeches by three key coup leaders: Colonel Assimi Goïta, Captain Ibrahim Traoré, and General Abdourahamane Tchiani. My analysis reveals four interconnected rhetorical frameworks: framing interventions as emergency measures, cultivating savior personas, weaponizing anti-colonial grievances, and deploying anti-corruption narratives. A comparative analysis of …

Nutrient Transport And Transformation In Anaerobic Co-Digestion Of Crop Residues And Swine Manure, Shelby Stoner

Theses and Dissertations

Cover crops may improve soil quality, water quality, soil health, reduce soil fertilizer use, and increase nutrient availability to cash crops. The benefits of cover crops are discussed due to the below ground presence of cover crops, but the aboveground biomass can serve as a source of sustainable energy through anerobic digestion. The nutrients in the above ground biomass can also be recycled and reused through land application of digestion effluent. We have tested anaerobic co-digestion of different biomass feedstocks: cover crops, soybean stalk, and corn stover with swine manure using both small and pilot scale digesters. Nutrients, specifically NPK, …