114333 research outputs found
Sort by
Comparison of Recycled Polypropylene Sourced From Food and Nonfood Applications for Direct Food Contact Use
No full textPolypropylene (PP) is a widely used polymer in packaging applications, which is often landfilled as part of #3–7 bales from material recovery facilities (MRFs). With rising demand for postconsumer PP, and its high composition in #3–7 bales, MRFs are increasing PP sorting capacity. Therefore, assessing chemical safety and physical quality becomes critical as design parameters for new products. Contamination emerges from various sources, for example, manufacturing, recycling and the original intended use (food compared with nonfood applications). Analysing the physical properties and contaminants/residual additives in postconsumer recycled (PCR) PP can provide guidance to improve cleaning, separation processes and quality reducing landfilling. PCR PP from #3–7 bale was sorted into two categories: food and nonfood applications. Each PCR PP type was exposed to a simulated recycling procedure and collected at each unit operation (washing, moulding, etc.) for analysis. It was determined nonfood-application PCR PP contained higher quantitative phthalates, bisphenols and qualitative IAS/NIAS with inconsistent contaminant removal during washing. Nonfood-application PCR PP possessed lower viscosity and molecular weight (Mw) across all categories except Mw in the unwashed sample compared with food-grade PCR PP, likely due to contamination, polymer degradation or additive levels. This research identifies potential contaminants and additives in food- and nonfood-application PCR PP from MRFs. These findings emphasise proper sorting for safe food-grade feedstocks from postconsumer sources. These data indicate careful sortation is required to reduce exposure to chemicals of concern and unapproved additives in food packaging that comprised PCR PP.This article is published as Tumu, K., Vorst, K., Curtzwiler, G., Comparison of Recycled Polypropylene Sourced From Food and Nonfood Applications for Direct Food Contact Use. Packaging Technology and Science. 2026, Early View. https://doi.org/10.1002/pts.70059.Funding
This work was supported in part by the Polymer and Food Protection Consortium (PFPC) at Iowa State University Agriculture and Home Economics Experiment Station HATCH Project 04202 and the Institute for the Advancement of Food and Nutrition Sciences (IAFNS) through funding by the ILSI North America Food and Chemical Safety and Food Packaging Safety Committees. IAFNS is a nonprofit science organisation that pools funding from industry collaborators and advances science through in-kind and financial contributions from public and private sector participants
Providers of relief in distress: RAG-based LLMs as situation and intent-aware assistants
No full textIn high-stress humanitarian and mental health contexts, timely access to accurate, empathetic, and actionable information remains critically limited, especially for at-risk and underserved populations. This work introduces LLooMi, an open-source, retrieval-augmented generation (RAG) conversational agent designed to deliver trustworthy, emotionally attuned, and context-aware support across domains such as mental health crises, housing insecurity, medical emergencies, immigration, and food access. Leveraging large language models (LLMs) with structured prompting, LLooMi reformulates user queries into actionable intents, often implicit, emotionally charged, or vague. It then retrieves and grounds responses in a curated, domain-specific knowledge base, without storing personal user data, aligning with privacy-preserving and ethical AI design principles. LLooMi adopts an intent-aware architecture that adapts its tone, content, and level of detail based on the user's inferred psychological state and informational goals. This step enables delivering fast, directive responses in acute distress scenarios or longer, validation-oriented support when emotional reassurance is needed, emulating key facets of therapeutic communication. By integrating NLP-driven semantic retrieval, structured dialogue memory, and emotionally adaptive generation, LLooMi offers a novel approach to scalable, human-centered digital mental health interventions. Evaluation shows an average answer correctness (AC) of 92.4% and answer relevancy (AR) of 84.9%, with high scores in readability, perceived trust, and ease of use. These results suggest LLooMi's potential as a complementary NLP-based tool for mental health support in digital psychiatry and crisis care.This article is published as Nazar, Ahmad, Brianna R. Norman, Halle Northway, Abrahim Toutoungi, Emma Zatkalik, Gabriel Carlson, Ellery Sabado, Hamza Shawa, and Mohamed Y. Selim. "Providers of Relief in Distress: RAG-based LLMs as Situation and Intent-Aware Assistants." Frontiers in Artificial Intelligence 9: 1712596. doi: https://doi.org/10.3389/frai.2026.1712596
Boar sperm dysfunction at the cellular and molecular level
No full textSuccessful reproduction is indispensable for species preservation, and sperm dysfunction contributes significantly to low fertility in both animals and humans. A comprehensive understanding of sperm health is critical for successful fertilization, as disruptions in sperm cellular structure, molecular integrity, and cell regulation pathways can impair sperm fertilization potential. This dissertation integrates proteomics, deep learning-based morphology analysis, and ferroptosis-focused biomarker profiling to develop objective, mechanistic, and high-throughput tools for sperm health evaluation in boars.
Chapter 2 investigated proteome conservation across swine, bovine, bubaline, murine, and human spermatozoa using the protein Basic Local Alignment Search Tool (pBLAST) similarity analysis, and the Protein ANalysis THrough Evolutionary Relationships (PANTHER) functional classification system. Livestock species, particularly swine, demonstrated the highest similarity to the human sperm proteome, supporting their suitability as an appropriate animal model for comparative and translational studies for male reproductive biology.
Chapter 3 describes the development of convolutional neural network (CNN) models for automated sperm morphology assessment using Image-Based Flow Cytometry (IBFC). Five CNN architectures capable of classifying twelve to fourteen different sperm morphologies with high precision were developed. These optimized models were extensively validated using independent datasets in Chapter 4 that were withheld from the training model creation in Chapter 3. The models indicated high performance against manual classification results when compared using different statistical methods, confirming strong diagnostic reliability across most morphological categories.
Chapter 5 investigated ferroptosis, a regulated, iron-dependent form of cell death, as a mechanistic driver of sperm deterioration during liquid semen storage. Different biomarker indicators were used for spectral flow cytometry-based ferroptosis profiling of intracellular glutathione, ferrous iron (Fe2+), zinc (Zn2+) levels, plasma membrane integrity, mitochondrial function, and oxidative stress in sperm cells. The results indicate that ferroptosis inducers accelerate sperm damage, while inhibitors partially rescue biomarker profiles. Untreated semen also demonstrated time-dependent shifts in ferroptotic biomarkers consistent with spontaneous ferroptotic activation.
Together, the studies in this dissertation provide a multidimensional framework linking proteomic composition, AI-assisted morphology profiling, and ferroptosis-mediated cell death to analyze sperm function. This research opens new avenues for understanding and improving fertility by establishing ferroptosis as a quantifiable and meaningful type of sperm cell death. The study holds significant potential for advancement in artificial insemination programs, refining sperm preservation methods, and potentially translating into human reproductive medicine
Super Arrhenius temperature dependent viscosity due to liquid-liquid phase separation in the super-cooled Kob-Andersen model
No full textIn this paper, we introduce a new order parameter called the weighted coordination number (WCN) to study the liquid-liquid (LL) phase separation, using the temperature-dependent coarsening of the LL interface as a possible mechanism for the glass transition. The well-established glass-forming Kob-Andersen binary Lennard-Jones system is used for our studies. The gas-liquid binodal line is reconstructed using the WCNs, and the same approach is extended to study the liquid-liquid binodal line. Systems of various densities are instantaneously quenched from high to low temperatures where a liquid-liquid separation is observed. Densities and the composition of each liquid state are used to check the level rule, along with density and pressure profiles, demonstrating local equilibrium of liquid-liquid phase separation. The transition from the liquid-liquid phase separation in the supercooled region to the glass transition region is modeled by adopting a Markov Network Model to estimate the temperature dependent viscosity using liquid-liquid interfacial information from the classification.This is a preprint from Brickley, Jayme, and Xueyu Song. "Super Arrhenius temperature dependent viscosity due to liquid-liquid phase separation in the super-cooled Kob-Andersen model." arXiv preprint arXiv:2602.16060 (2026). doi: https://doi.org/10.48550/arXiv.2602.16060.This work is supported by the Division of Chemical and Biological Sciences, Office of Basic Energy Sciences, U.S. Department of Energy, under Contact No. DE-AC02-07CH11358 with Iowa State University
Incorporating structure of data collection in analysis: Analysis of tensor-variate and massive spatial data
No full textThis thesis consists of two major parts centered on developing novel statistical methodologies for analyzing massive high-dimensional, structured data, particularly tensor-variate and spatially indexed functional data, motivated by real-world applications in neuroscience and environmental science.
In the first part, we focus on tensor-on-tensor time series regression (TOTTR), designed for the analysis of massive array- or tensor-variate data, such as those arising from functional magnetic resonance imaging (fMRI). Although such data may appear to be a straightforward extension of multivariate data, they often present significant computational, methodological, and theoretical challenges. Traditional fMRI studies often employ a two-stage pipeline: voxelwise univariate time series regression followed by spatial modeling of summary statistics, which may lead to information loss and inefficiencies. We propose a holistic one-step tensor regression framework that models the tensor-valued time series response with a possible tensor-variate covariate, leveraging low-rank structures of the regression coefficient tensor and Kronecker separable covariance structure to ensure scalability. Parameter estimation is performed using the alternating least squares method, and the approach is implemented in a computationally efficient R package that incorporates numerous simplifications for speed and scalability, and optionally includes a Ridge penalty in the optimization process. We apply the method to both simulated datasets and a multi-subject fMRI study on Major Depressive Disorder (MDD), successfully identifying significant brain activation patterns.
We further establish a theoretical consistency of our estimator using perturbation bounds, a type of result that is sparse in the literature and much harder to obtain than its matrix counterparts. This theory, in turn, also yields a related estimation procedure that is often faster.
In the second part, we turn to modeling multivariate spatial and spatially indexed functional data. For multivariate spatial data, we develop a matrix-free estimation framework based on factor models. The factor scores are modeled using lattice approximations of fractional Gaussian fields, which encompass intrinsic random fields and a broad class of spatial models. This approach flexibly captures anisotropy and long-range dependence and is scalable to large and high-dimensional, potentially misaligned datasets via a stochastic EM algorithm. We demonstrate the methodology using groundwater quality data from Bangladesh, where the British Geological Survey and Bangladesh Department of Public Health Engineering collected measurements of many minerals at each location, with particular concern about arsenic contamination in groundwater.
Spatially indexed functional data, i.e., spatially spread out data points that are themselves observations of functions defined over a continuous variable, are becoming increasingly common. However, methods to analyze functional data with spatial dependence are relatively underdeveloped. We develop a semi-parametric flavored modeling framework that combines the aforementioned spatial discretization, factor model-based dimension reduction, and spline-based functional regression. The resulting method is scalable, interpretable, and matrix-free, making it suitable for large-scale oceanographic datasets. We illustrate its performance using Argo-float data, where temperature profiles are observed across ocean locations and depths.
Together, the methods developed in this thesis form a unified and computationally efficient toolkit for analyzing complex structured datasets in neuroscience, geostatistics, and environmental monitoring, while also advancing the theoretical understanding behind these models
Design and test techniques to enhance analog and mixed-signal (AMS) circuits performance and reliability
No full textThere has been a significant rise in the demand for integrated circuits (ICs), driven by their wide-ranging applications in industrial automation, automotive systems, the Internet of Things (IoT), and emerging technologies such as Artificial Intelligence (AI) and Machine Learning (ML). This surge in demand is coupled with an industry-wide push for smaller chip designs that maintain or even enhance performance metrics. At the same time, as semiconductor chips
integrate more functionality, reliability and safety concerns have become central to mitigating failures in them. The research work outlined in this dissertation encompasses various analog and mixed-signal (AMS) design and test techniques that cater to the pressing needs of the industry, “achieving IC performance reliably”.
Precision analog design is based on device matching and layout is key to achieving proper matching. This is because AMS circuits are often governed by ratio metric properties. As a result, their performance largely depends on differential variability between adjacent devices on the same die or wafer. In this dissertation, we propose a set of techniques/algorithms to generate layout patterns that match both linear and nonlinear gradient effects between critical circuit components for different analog circuit classes and applications.
The growing IC demand is accompanied by rising expectations for defect-free systems. Publicly available systems-on-chip (SoC) data indicates that analog and mixed-signal (AMS) circuits contribute to over 80% of chip failures. This dissertation addresses the growing need introducing digitally assisted defect detection methods for analog circuits. Several chapters of this dissertation are dedicated to the development of built-in self-test (BIST) defect detection techniques specifically focusing on operational amplifiers (op amps), phase-locked loops (PLLs) and R-2R digital to analog converters (DACs).
As semiconductor technology continues to scale into deep-submicron processes, supply voltages and available chip area for analog designs have been steadily reduced. While this scaling trend has enabled higher integration and improved digital performance, it poses significant challenges for analog circuits, whose operation is inherently sensitive to voltage headroom, device area, and power. To address these challenges, this dissertation presents compact, low-power design
techniques that achieve improved performance in key analog building blocks, including voltage
reference circuits, R-2R digital-to-analog converters (DACs), and low-dropout regulators (LDOs)
Structure-aware graph representation learning using graph neural networks
No full textGraph Neural Networks (GNNs) have emerged as powerful tools for learning representations from graph-structured data across diverse domains. However, existing GNN architectures often fail to adequately capture critical structural properties that fundamentally influence both the quality and fairness of learned representations. This dissertation addresses three fundamental challenges in structure-aware graph representation learning: preserving directional information in directed graphs, ensuring structural fairness in node representations, and mitigating community-level structural bias in GNN predictions.
First, we propose a novel method for computing joint two-node structural representations for link prediction in directed graphs. Existing approaches either learn node embeddings independently and combine them, failing to differentiate distant nodes with similar neighborhoods, or employ undirected GNNs on enclosing subgraphs, inevitably losing directional signals. Our approach utilizes directed enclosing subgraphs with direction-aware positional encodings and GNNs to preserve edge orientation, demonstrating superior link prediction performance compared to both undirected and existing directed baselines.
Second, we address individual structural bias in GNN-based graph representation learning by incorporating both local and global structural information into representation learning. Prior fairness work has focused primarily on node features while overlooking structural biases that can systematically disadvantage individuals based on their position in the graph. We propose a pre-processing bias mitigation approach that employs locally fair PageRank methods to address local structure discrepancies and truncated singular value decomposition-based similarities to handle global structural disparities between node pairs. This method achieves superior individual fairness metrics while maintaining predictive performance.
Third, we address community-level structural bias in GNN-based graph representation learning, which arises from diverse local neighborhood distributions during GNN message passing. Current GNN fairness research relies on oversimplified evaluation metrics that can provide misleading assessments of fairness. We introduce ComFairGNN, a novel framework that measures and mitigates bias at the community level using a learnable coreset-based debiasing function. This approach addresses the complex evaluation paradoxes inherent in graph-structured data and demonstrates effectiveness across both accuracy and fairness metrics.
Comprehensive evaluations on multiple benchmark datasets validate that our structure-aware approaches significantly outperform state-of-the-art baselines in their respective tasks. This dissertation establishes that explicit modeling of structural properties, including directionality, positional context, and community-level patterns, is essential for developing GNN architectures that are both effective and equitable for real-world applications
Preparing future teachers for AI-enhanced science classrooms: Insights from a study of elementary teacher candidates
No full textGrounded in qualitative data from 30 elementary teacher candidates (ETCs) in the Midwestern U.S., this case study examines ETCs’ readiness for AI-enhanced classrooms and finds that ETCs hold moderate to low confidence in teaching biology, yet express a desire to teach it more engaging for young learners. Although many ETCs had negative or vague prior perceptions of Artificial Intelligence in education (AIEd), guided classroom experiences helped them recognize its potential and limitations. Despite this growth, ETCs continued to demonstrate gaps in AI ethics and technical understanding. The study identifies ETCs’ competencies at an early developmental stage, recommends increased exposure to authentic AIEd applications, scaffolded instruction on AI foundations and ethics, and participation in professional learning communities. These insights contribute to how ETCs conceptualize and adapt to emerging technologies, offering implications for curriculum design and policy development in teacher education.This accepted article is published as Jiang, A.H., Bahng, E.J., Coffman, C., Shelley, M., Gilbert, S., Fieffer, S., Abudagga, O., Preparing future teachers for AI-enhanced science classrooms: Insights from a study of elementary teacher candidates. Journal of Digital Learning in Teaching Education. January 2026. 1–12. https://doi.org/10.1080/21532974.2025.2606665.Funding: This work was supported by the Center for Excellence in Learning and Teaching (CELT) at Iowa State University, United States under the Miller AI Initiative
Developing a nanoscale force-mediated drug delivery system for lung fibrosis
No full textLung fibrosis is a disease characterized by high amounts of cell-generated forces occurring in a pro-fibrotic feedback loop. These forces exacerbate inflammatory responses from immune cells, as well as encouraging cell migration towards the extracellular matrix fibrils in fibrotic tissue. There is a lack of drug delivery systems capable of delivering drugs in response to cellular level forces, particularly in the extracellular matrix. In order to create a drug delivery system there needs to be a structure capable of securing cargo until release, a release mechanism that opens the structure based on cellular forces exerted on the extracellular matrix, and a way to attach the drug delivery system to the extracellular matrix. In this dissertation, I designed a DNA origami structure capable of being loaded with small molecules and proteins. I developed a method of selecting DNA aptamers that are highly specific towards fibril structure. I also characterized DNA force sensors based on their GC% and rupture method. Combining these three aspects, I will design a nanoscale, force-mediated drug delivery system
Influence of climate variability on the hydromechanical behavior of earthen dams
No full textEarthen dams constitute one of the most widely used forms of hydraulic infrastructure, yet their long-term performance is increasingly threatened by aging materials and growing climate variability. Intensifying precipitation patterns, prolonged wet periods, and fluctuating reservoir levels have amplified seepage‐related risks, which remain among the primary mechanisms of distress and failure in these structures. Despite substantial research on dam failure modes and hydrologic hazards, significant gaps persist in understanding how nonstationary climatic forcing interacts with the hydraulic behavior of aging earthen dams. Current engineering practice continues to rely on stationary assumptions and simplified boundary conditions, limiting the ability to assess future seepage evolution under changing environmental conditions.
To address these gaps, this thesis employs a comprehensive, multi-layered methodology encompassing a literature review, climate analysis, numerical modeling, and case-study validation. The literature review synthesizes research on climate variability, dam failure mechanisms, and numerical approaches for hydromechanical analysis, revealing fragmented treatment of climate-soil interactions and limited integration of stochastic climate processes into seepage evaluation. Building on this foundation, the research develops a climate-informed numerical framework that couples a nonstationary stochastic precipitation generator with a land–climate boundary interface to translate precipitation, evaporation, and reservoir fluctuations into transient hydraulic loading conditions. This framework is implemented through finite-element seepage modeling and calibrated using historical piezometer and reservoir records from Fontenelle Dam, an aging zoned earthfill structure with a documented history of seepage.
The findings demonstrate that climate-adjusted simulations produce pore-water pressures and saturation patterns more variable than those predicted under traditional stationary analyses. Extended wet periods and elevated reservoir stages were shown to drive disproportionate increases in hydraulic gradients, highlighting the sensitivity of aging dams to evolving climatic regimes. The results emphasize the importance of accounting for climate nonstationarity when evaluating seepage performance, as stationary assumptions systematically underpredict hydraulic demand and may obscure emerging risks in deteriorating infrastructure.
In bringing together climate modeling, unsaturated flow processes, and geotechnical analysis, this thesis establishes a practical framework that can be used to evaluate how future climatic conditions influence seepage in earthen dams. The results deepen our understanding of how aging materials and shifting hydroclimatic patterns interact within these structures, offering a clear pathway for incorporating climate considerations into dam-safety assessments. The framework developed here also creates a foundation for expanding future work toward fully coupled hydromechanical modeling and risk-informed decision tools. Collectively, these contributions support more resilient and forward-looking management of earthen dam infrastructure as climatic conditions continue to evolve