31825 research outputs found
Sort by
Stochastic resonance stimulation effect on stability during walking in people with Parkinson disease
No full textThis article was originally published in International Biomechanics . The version of record is available at: https://doi.org/10.1080/23335432.2025.2584665
© 2025 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits
unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow
the posting of the Accepted Manuscript in a repository by the author(s) or with their consent.People with Parkinson disease (PwPD) often face challenges with maintaining balance while walking, which can stem from sensory dysfunction. Studies have identified different biomechanical strategies that aid in preserving upright balance control. Stochastic resonance (SR) stimulation delivers sub-threshold electrical noise to enhance the detection capabilities of dysfunctional sensory systems. Yet, the effectiveness of SR in enhancing gait stability in PwPD is undetermined. The purpose of this study was to investigate the effects of SR on balance control during visually perturbed walking in PwPD (NCT06829342).
Fourteen individuals with PD completed the study. We established individualized sensory thresholds for SR stimulation and identified the optimal SR intensity. Following this, the participants walked within a virtually perturbed environment. Center of mass (CoM) excursion, foot placement, and ankle roll responses were assessed bilaterally.
Peak CoM excursion showed a significant increase, indicating reduced stability, with the SR condition compared to no-SR at the more affected side. Outcome measures related to balance control mechanisms were insignificant.
With SR, PwPD were driven by the induced fall with more sway and without significant alterations in balance strategies, which might be due to adding more noise to sensory processing and misidentifying the more affected side.This work was supported by the Parkinson’s Foundation (PF-JFA-2036) under the Stanley Fahn Junior Faculty Award given to Hendrik Reimann. Any opinions, findings, and conclusions expressed in this material are those of the authors. This funding source played no role in the design of this study, analysis, interpretation of data, or manuscript writing
Confronting the privacy leak epidemic in the machine learning era
No full textYuan, XuAs machine learning becomes increasingly integrated into modern digital infrastructure and mobile applications, concerns about user data privacy have grown significantly. Advanced ML models frequently rely on sensitive personal data to deliver intelligent and personalized services. However, this reliance also introduces serious risks: user data may be collected without consent, misused during model training, or leaked through model predictions. Addressing these threats requires a comprehensive understanding of how privacy risks manifest across the machine learning pipeline. This dissertation presents a systematic investigation into three critical dimensions of privacy in ML systems. First, we identify and characterize privacy leakage sources in both application-layer services and model behaviors. We analyze the mobile notification ecosystem as an overlooked but pervasive channel for covert data harvesting, and we introduce a novel self-comparison membership inference attack to expose how trained models reveal information about their training datasets. Second, we develop mechanisms for detecting unauthorized data usage. We propose new inference-based auditing techniques for semi-supervised models and introduce a non-intrusive, information-theoretic framework for dataset-level auditing in already trained models. Third, we design defense strategies to prevent privacy leakage, focusing on link inference attacks in Graph Neural Networks. We develop a structure-aware defense that obfuscates graph topology while preserving model utility. Collectively, this dissertation offers a unified view of data privacy risks in ML services. It contributes both empirical techniques and theoretical foundations for identifying leakage, detecting misuse, and defending against privacy threats, laying the groundwork for building secure and accountable machine learning systems.University of Delaware, Department of Computer and Information SciencesPh.D
Romantic relationships during emerging adulthood: associations with adult attachment and previous relationship experiences
No full textGoldstein, SaraThe current study explores the attachment styles and characteristics of individuals who have been in romantic relationships. This research expands previous romantic relationship experiences to include online dating, including topics of online grooming and dating app usage. Participants (n = 210) completed an online questionnaire assessing their attachment, relationship quality, childhood experiences, and online experiences. Analyses focused on examining associations among study variables. In general, results were consistent in that those participants who had insecure attachment tendencies also had more challenges in their romantic relationships, both in person and online. Specifically, insecure attachment was associated with reduced relationship quality and increased relationship difficulties, as well as with negative experiences online. Implications of these findings are discussed, and ideas for future research are explored.University of Delaware, Department of Human Development and Family SciencesM.S
Delaware 2013-2023 Behavioral Risk Factor Survey
No full textThis report analyzes the 2013 and 2023 BRFSS data. The objective of BRFSS is “to collect uniform state-specific data on health risk, behaviors, chronic diseases and conditions, access to health care, and use of preventive health services related to the leading causes of death and disability in the United States”. The goal of the report was to present the BRFSS survey results and provide statistical analysis of the Delawareans’ health as well as their health behavior patterns while comparing data from two surveys conducted in 2013 and 2023. The 2013 BRFSS combined landline and cell phone weighted response rate for Delaware was 5,343; whereas the 2023 BRFSS combined landline and cell phone weighted response rate for Delaware was 4,427. To make a meaningful comparative analysis of the results of two different BRFSS survey separated by ten years, only those sections and modules were selected in the 2023 survey that had corresponding sections and modules in the 2013 survey. This study generalizes the prevalence rates of health issues, health care access, chronic diseases, and some behavioral risk factors among the population
Magnitude, concentration, and metric complexity in phylogenetics and information theory
No full textDinh, VuMadiman, MokshayThe results in this thesis contribute to two areas: phylogenetic inference and information theory. In the first part, we investigate some probabilistic models and problems that arise when one wants to estimate evolutionary parameters from large biological datasets. The classical setting for inference assumes that the number of independent sites, k, is much larger than the number of leaves, n, on a phylogenetic tree. In Chapter 2, we add to the list of examples of how estimation can go wrong when k is much smaller than n. We show that on the balanced claw tree under the binary non-symmetric model, the maximum likelihood estimator of evolutionary parameters is inconsistent when k = 1. ☐ To prove results about consistency in the classical case, concentration inequalities for independent random variables are used. However, when k << n, the dependencies between the observations cannot be ignored. So in Chapter 3, we show some McDiarmid-type concentration inequalities for leaf marginals of tree-indexed Markov measures and Bayesian Networks, which might be of interest to the machine learning community as well. ☐ Chapters 4 forms the bridge from phylogenetics to information theory, via the problem of ancestral state reconstruction. The existence of consistent estimators of the root state for discrete models is known to be equivalent to a density assumption near the root, called the Big Bang condition [12]. In the continuous case, the consistency of the Maximum Likelihood Estimate (MLE) based on observations Yn from a Brownian Motion (BM) model is known to be equivalent to the condition Magnitude (Vn) → +∞ where Vn is the covariance matrix of Yn. These geometric properties were shown to be equivalent in [8] using probabilistic tools. In this chapter, we reprove this equivalence in the language of magnitude [1], and lay a geometric framework for maximum likelihood estimation of the root state. ☐ The magnitude of a finite metric space is a quantification of its “effective” size [1]. An open problem in the theory of magnitude and diversity is that of finding an operational meaning for these quantities in information theory [4]. In the theory of lossless coding, a fundamental problem is that of quantifying the size of the smallest possible set of n-letter words from a discrete memoryless source with the largest possible likelihood. This establishes the lowest possible “cost” to be paid in order to communicate or compress n-sized strings of data with least possible error. The Asymptotic Equipartition Property (AEP) provides a concrete candidate for such a set, namely, the strongly or weakly typical sets [5]. These sets also give us an entry point for metric generalizations of the lossless coding problem to letters with a notion of distance between them. ☐ In Chapter 5, we lay the foundations for a metric generalization of the AEP in an effort to give operational meanings to magnitude and diversity. We show that for alphabets with two letters, the magnitude of type classes is approximately 2nHK(p), where HK(p) is the metric 1-complexity, which is the logarithm of the diversity of order 1. Methods to extend these results to alphabets of any size are also described, and in Chapter 6, we outline the requirements to prove a metric generalisation of the AEP, which will be completed in future work.University of Delaware, Department of Mathematical SciencesPh.D
Life Cycle Assessment of Thermoelectrics: Ecological Viability in Intermittent Waste Heat Scenarios
No full textThis article was originally published in ACS Omega. The version of record is available at: https://doi.org/10.1021/acsomega.4c08192.
Copyright © 2025 The Authors. Published by American Chemical Society. This publication is licensed under CC-BY-NC-ND 4.0 (https://creativecommons.org/licenses/by-nc-nd/4.0/).This study evaluates the ecological impacts of thermoelectrics (TEs) in stationary applications that periodically generate waste heat using life cycle assessment (LCA) methodology, a first of its kind. Six TE modules are analyzed for a periodic heat-emitting application: a natural gas-based power plant that meets only peak electricity demand. The analysis uses detailed inventories from an earlier study regarding the production and end-of-life stages of the TEs. The results show that while TEs are effective in conserving fossil fuels and lowering greenhouse gas emissions, they do not exhibit significant positive effects on other environmental impacts. These findings persist even when accounting for variations in TE conversion efficiency and lifetime and the implementation of a circular economy approach for recycling and repurposing TE modules. This suggests that the environmental suitability of TEs is predominantly influenced by the type of fossil energy source they replace, making current TEs unsuitable for stationary applications that periodically generate waste heat. The study also highlights the need for further research on the development of new, practical TEs that utilize nontoxic, abundant elements and are produced through less energy-intensive techniques.The authors declare that there are no acknowledgments to report for this manuscript
Checking for understanding: administrator-led efforts for developing teachers to deliver evidence-informed reading instruction
No full textVanGronigen, Bryan A.There is a century-old debate surrounding reading instruction that has been termed “The Reading Wars” (Kim, 2008). In 2022, the Pennsylvania General Assembly passed Act 55, a piece of legislation focused on school districts providing their teachers with professional development on the Structured Literacy (SL) approach to reading instruction. This Education Leadership Portfolio (ELP) aimed to address the problem of teachers having too little understanding of reading development by enhancing teachers’ understanding of reading development and, ultimately, their reading instructional practice. Three improvement strategies were enacted to foster this enhancement: (a) understand teachers’ knowledge of reading development, (b) design and pilot a professional development series to enhance teachers’ reading instructional practice using the Science of Reading (SoR) principles and the SL approach, and (c) evaluate the initial implementation of the professional development series pilot. Data collection included surveys, observations, and interviews to learn about what informs teachers’ reading instruction, their understanding of reading development, and factors that enable or hinder teachers’ ability to implement evidence-informed instruction in their classrooms. Results suggested that teachers had increased awareness of SoR principles that undergird the SL approach, which illustrated a shift toward data-driven systematic and explicit reading instruction. Results also demonstrated increased student achievement in foundational skills based on Pennsylvania assessments. Overall, this ELP underscored the importance of an administrator’s role as an instructional leader in creating conditions to enhance teachers’ instructional practice, particularly evidence-informed reading instruction. This ELP has implications for several constituent groups. First, an elementary school should support teachers’ reading instructional practice, specifically language comprehension instruction and Duke and Cartwright’s (2021) research on the active view of reading to enhance teachers’ reading instructional practice. Next, as principals implement Act 55, they—themselves—need to work to understand their teachers’ understanding of reading development by gathering and analyzing teachers’ and support staff members’ perceptions. By understanding these perceptions, principals can leverage professional development to enhance teachers’ reading instructional practice. Finally, researchers are urged to investigate the long-term effects of teachers’ reading instructional practice that include SL to examine influences on students and their outcomes (e.g., if fewer students are identified for special education as a result of SL implementation in early grades). Such investigations should include urban, suburban, and rural schools to account for contextual differences. This kind of research builds upon this ELP to consider how more students across these contexts can have more equitable access to high-quality reading instruction.University of Delaware, School of EducationD.Ed