21794 research outputs found
Sort by
Racial/ethnic Differences In The Rates Of Subjective Cognitive Decline And Related Outcomes
No full textSubjective cognitive decline (SCD) is a precursor to Alzheimer’s disease and related dementias (ADRD), which is projected to more than double in prevalence by 2060. Rates of SCD and ADRD are especially pronounced in Hispanic and non-Hispanic Black (NHB) Americans when compared to non-Hispanic Whites (NHWs). Therefore, this study is focused on understanding how risk factors for ADRD predispose racial/ethnic (R/E) minorities for SCD and SCD-related functional limitations. Using the Center for Disease Control and Prevention (CDC)’s Behavioral Risk Factor Surveillance System (BRFSS) data, I examined risk factors for SCD and SCD-related outcomes in NHBs, NHWs, and Hispanics. Overall, there were higher rates of SCD-related functional limitations in the overall sample and a subsample of individuals with SCD and vascular conditions. Additionally, I found that NHBs and Hispanics were more likely than NHWs to belong to more severe SCD-related latent classes. My findings point to the need for policy changes regarding the availability of protective factors, such as education, safe housing, health insurance, etc., especially for more susceptible R/E groups. Further research is needed to understand the relationship between R/E minorities and their healthcare providers to provide insight to the decreased use of healthcare services by R/E minority groups than NHWs
A Unified Deep Learning Framework For Simultaneous Segmentation And Multiclass Classification Of Chronic Wounds *
No full textChronic wounds affect millions worldwide, posing significant challenges for healthcare systems and a heavy economic burden globally. The segmentation and classification (S&C) of chronic wounds are critical for wound care management and diagnosis, aiding clinicians in selecting appropriate treatments. Existing approaches have utilized either traditional machine learning or deep learning methods for S&C. However, most focus on binary classification, with few addressing multi-class classification, often showing degraded performance for pressure and diabetic wounds. Wound segmentation has been largely limited to foot ulcer images, and there is no unified diagnostic tool for both S&C tasks. To address these gaps, we developed a unified approach that performs S&C simultaneously. For segmentation, we proposed Attention-Dense-UNet (Att-d-UNet), and for classification, we introduced a feature concatenation-based method. Our framework segments wound images using Att-d-UNet, followed by classification into one of the wound types using our proposed method. We evaluated our models on publicly available wound classification datasets (AZH and Medetec) and segmentation datasets (FUSeg and AZH). To test our unified approach, we extended wound classification datasets by generating segmentation masks for Medetec and AZH images. The proposed unified approach achieved 90% accuracy and an 86.55% dice score on the Medetec dataset and 81% accuracy and an 86.53% dice score on the AZH dataset These results demonstrate the effectiveness of our separate models and unified approach for wound S&C.
Wound segmentation aids in the measurement of wound area which further assists in analyzing the wound healing progress. In this research work, we have also presented a deep learning-based segmentation approach namely Dual-UNet for precisely segmenting the diabetic foot ulcers images. The foot ulcer images are passed to Dual-UNet to generate the mask image having the segmented wound area. In our proposed approach, two UNets are utilized each having encoder, atrous spatial pyramid pooling (ASPP) block, decoder with skip connections, and output block to improve the segmentation results. We assessed our framework by performing the experimentation on two benchmark datasets named foot ulcer segmentation (FUSeg) challenge and AZH segmentation dataset. Additionally, we have also evaluated our Dual-UNet for cross-dataset validation to demonstrate the generalization capability of the proposed approach. Both the quantitative and visual results demonstrate the effectiveness of the proposed framework for the segmentation of chronic diabetic wounds.
We also propose a novel lightweight fused-densenet method capable of reliable classification of multiple types of chronic wounds. Our method comprises a fully trained and a partially trained densenet model, which is fused to develop an effective multiclass wound classification approach. We introduce the GeLU activation function to tackle the dying ReLU problem, enhanced performance, better learning, and efficient training. Further, we add the dense and dropout layers along with the L2 regularization approach to counter the model overfitting. We assessed the performance of our lightweight model on the standard Medetec and AZH datasets, as well as their augmented versions. We employed multiple augmentation techniques to increase the number of samples and diversity of these datasets to tackle the overfitting and class imbalance issues. Experimental evaluation on AZH, Medetec, and augmented versions of both datasets signifies the efficacy of our proposed method for multiclass wound classification
Compensatory Roles Of Jaks And Stats In Mammary Gland Development And Breast Cancer
No full textJanus kinases (JAK) and Signal Transducers and Activators of Transcription (STATs) constitute a central signaling module relaying hormone and cytokine cues and instructing a multitude of pro-tumorigenic processes such as proliferation, differentiation, and invasion in breast cancers. Previous studies have demonstrated that JAK1 and JAK2 have discrete biological functions in pregnancy-associated mammary gland development and breast tumorigenesis, mediated by specific STAT transcription factors, in particular STAT3, STAT5a, and STAT5b. However, the cooperative and compensatory actions of JAKs and STATs and their effects on early mammary gland development and breast cancer initiation and progression remain largely unknown. Herein, we report that female mice with a mammary epithelial-specific deletion of both JAK1 and JAK2 exhibit a severe block of the development of mammary gland ducts that persists through adulthood. Mammary transplantation experiments confirmed that the essential functions of JAK1 and JAK2 in ductal elongation are mammary-epithelial cell intrinsic. A comparative analysis of mice with combinatorial deletions of Jak1 and Jak2 alleles suggests that JAK2 is the main Janus kinase regulating early mammary ductal elongation and that one copy of Jak1 can compensate for JAK2 deficiency. After the generation and assessment of quadruple STAT1/3/5a/5b mammary glands, we found that the combined deletion of STATs was not sufficient to block ductal development despite the lack of activation of all seven STATs. These results suggest that in addition to their discrete STAT-dependent functions during the gestational cycle, JAK2 and JAK1 have STAT-independent non-canonical functions that are essential for mammary gland ductal elongation. In this study, we also explored JAK/STAT signaling in two different murine triple-negative breast cancer models. Using a claudin-low mammary tumor model driven by mutant KRASG12D expression, we show that (1) JAK1 mediates the oncogenic activation of STAT3 in claudin-low breast cancers, (2) the simultaneous deletion of STAT3 and STAT5 prevents mammary tumor initiation, and (3) the loss of STAT3/5 in claudin-low mammary tumors leads to the same compensatory activation of STAT1 that we identified in the normal mammary epithelium. Additionally, we generated a murine model that overexpresses TSG101 leading to the development of triple-negative adenosquamous carcinomas. We also demonstrated that the perpetual expression of TSG101 is essential for cancer cell survival in established tumors. Moreover, we found that adenosquamous carcinomas show enriched signatures of cytokine interaction and JAK/STAT signaling. On the molecular level, these tumors expressed high activated levels of STAT3, STAT5, and an inactivation of STAT1. In summary, our studies uncovered novel cooperative and compensatory functions of JAKs and STATs proteins in both mammary gland development and breast cancer. We generated the first mouse model that is completely deficient in the expression and activation of all known STAT proteins. We also made the paradigm-shifting observation that important roles of JAK1/2 are mediated by noncanonical mechanisms that may or may not require their tyrosine kinase functions. The collective results of our experimental models also provide evidence that the co-targeting of STAT5 and STAT3, and potentially their upstream activators JAK1 and JAK2, may serve as a preventive or therapeutic strategy for the highly aggressive triple-negative breast cancer subtype
Navigating Postsecondary Education: Exploring The Impact Of Dual Enrollment On Rural Students’ Postsecondary Pathways
No full textThis dissertation investigates the impact of dual enrollment participation on postsecondary institutional choice and persistence among rural students, using data from the Beginning Postsecondary Students Longitudinal Study (BPS:12/17). While dual enrollment programs are associated with positive postsecondary outcomes for general student populations, limited research explores their effects on rural students, who face unique barriers such as geographic isolation and limited access to advanced coursework. Guided by Tinto’s Theory of Social and Academic Integration, this study addresses two research questions: (1) How do postsecondary institutions chosen by rural dual enrollment participants compare to those chosen by non-participants? (2) How does persistence through year three differ between rural students who participated in dual enrollment and those who did not?
Using logistic regression, this study found that rural students who participated in dual enrollment were more likely to enroll in 4-year postsecondary institutions rather than 2-year institutions, though this association was not statistically significant. However, dual enrollment participation showed a statistically significant relationship with persistence, indicating that rural students who earned dual enrollment credits were more likely to remain enrolled through their third year of college or complete their degree. These findings highlight dual enrollment’s role in fostering long-term college persistence among rural students.
This research highlights the complexity of dual enrollment’s influence on rural students and emphasizes the importance of program context. The absence of data on dual enrollment settings (e.g., a college campus versus a high school or online) limits the understanding of the role of social and academic integration. These findings underscore the need for policymakers and educators to tailor dual enrollment programs to foster academic success and belonging, ensuring rural students achieve long-term postsecondary outcomes
Evaluation Of Metal Homeostasis In Select Biological Systems
No full textMetals play an indispensable role in biology. In eukaryotes, disruption of metal homeostasis, either by genetics leading to metal overload or exposure to toxic heavy metals, has fatal health outcomes. Metal chelation therapy is the primary treatment for these disorders to facilitate elimination of excess or toxic metals. However, many of these chelators are met with limited efficacy due to their poor selectivity between the target metal and essential physiological metals. This is compounded by their tight binding affinities, resulting in disruption of metal homeostasis and essential cellular functions. A major focus of this work is characterizing the novel iron chelator, ATH434, previously shown to have promising results reducing iron overload within in vivo animal models and in clinical trials.
Here, our biophysical studies of ATH434 indicate the compound coordinates Fe2+ using oxygen and nitrogen ligands in a 1:1 stoichiometric ratio through partial coordination with water. Additionally, ATH434’s sub micromolar affinity for Fe2+ and Fe3+ poises ATH434 to sequester excess Fe2+ with affinity values and a coordination architecture that mimics endogenous iron chaperones. Further analyses of ATH434 indicate that, in addition to its therapeutic potential, the drug also behaves as a metal responsive fluorophore capable of detecting in vitro transition metal binding. Here we show ATH434 also binds other transition metals (Co2+, Ni2+, Cu2+, Zn2+, and Cd2+) with low micromolar affinity. Our studies demonstrate ATH434 is a tool compound compatible with high-throughput screening and we demonstrate the application of ATH434 in identifying a novel Cd specific peptide, HSQKVF. Metals also play an essential role in other biological processes, including prokaryotic metabolism and survival, as well as environmental applications where they can serve as catalysts for sustainable energy solutions. To investigate the role of metals in prokaryotic systems, we’ve expanded our evaluation of metals in biology to characterize the role of the bacterial protein Fpa within the mammalian pathogen Staphylococcus aureus. Using X-ray absorption spectroscopy and metal binding assays, we demonstrate Fpa is an iron binding protein consistent with its physiological role in Fe homeostasis. Finally, we’ve further broadened our focus to demonstrate the function of Ni as a catalyst within a de novo metalloprotein that can act as an alternative energy source by facilitating Ni mediated photocatalytic H2 evolution, as a clean and sustainable fuel source. Our studies indicate the peptide oligomer, 4SCC, exhibits photocatalytic activity in the Ni2+ bound form and we identify the catalytic metal binding site composed of four sulfur ligands via X-ray absorption spectroscopy. Overall, this work highlights the versatility of metals in biological systems and emphasizes the importance of continued investigation to identify the function of metals in fundamental biological processes and disease
Mass Spectrometry Combined With Chemical Biology To Study Human Ailments
No full textPost-translational modifications (PTMs) are essential as chemical changes that regulate protein function, activity, stability, interactions, and localization after protein synthesis. PTMs, such as phosphorylation, acetylation, and ubiquitination, can activate or deactivate proteins, influence their interactions with other molecules, and determine their cellular location. This regulation is vital for numerous biological processes, such as cell cycle control, signal transduction, and gene expression. Abnormal PTMs are linked to various diseases in humans, including parasitic infection, blood disorders, cancers, and neurodegenerative diseases, making their study crucial for understanding disease mechanisms and developing targeted therapies. In the Pflum lab, we are interested in the PTMs in cell biology to understand the molecular basis of disease. We are mainly interested in two proteins that regulate PTMs involved in human diseases: kinase and histone deacetylase enzymes. The current gap in the field for both kinases and HDAC is the lack of tools available to identify both the protein substrates and associated proteins. Our goal is to use a chemical approach combined with LC-MS/MS to characterize the substrates of kinases and HDAC proteins in human aliments, which will lead to the identification of novel drug targeting and biomarkers to diagnose illness. To improve the gap in the field for studying kinases, the Pflum lab has developed kinase-catalyzed labeling and crosslinking techniques to identify kinase-substrate pairs. The methods utilize γ- modified ATP analogs that contain crosslinking analogies, which help identify kinase-substrate pairs by labeling substrates or stabilizing the kinase-substrate interactions through crosslinking. K-CLASP (Kinase Catalyzed Crosslinking and Streptavidin Purification) is a method for identification of phosphosite-specific kinase and unanticipated protein-protein interactions in phosphorylation-mediated biological events. K-CLASP was successfully applied to Plasmodium falciparum in two collaborative studies with Dr. Lanzer\u27s group at Heidelberg University Hospital, which discovered kinases and protein-protein interactions. The data provided can lead to improved development of therapeutics for malaria. Lastly, a new and state-of-the-art method was created in the Pflum to study Histone deacetylase1 (HDAC1) proteins, which are viable targets for future sickle cell disease (SCD) drug development. HDAC1 inhibitors influence globin switching. Unfortunately, the limited understanding of the role of HDAC1 in sickle cell biology and globin switching has stalled efforts to develop HDAC-targeted drugs for SCD symptoms. HDAC proteins catalyze the deacetylation of acetyl-lysine residues on HDAC proteins and are epigenetic regulators of gene expression. However, their role in regulating cellular activities beyond gene expression has grown. Here, we identified HDAC1 substrates in primary erythroblasts from sickle cell patients to characterize HDAC1 activity using substrate-trapping and LC/MS/MS. A requirement of substrate trapping is the transient expression of mutant HDAC1 in the cell line of interest. Due to poor transfection efficiency, overexpressing inactive mutants of HDAC1 in primary erythroblast cells is challenging and prevents direct trapping. To overcome the transfection challenge, we developed a double immunoprecipitation (dIP) trapping method to identify substrates and characterize the activity of HDAC1 proteins in SCD. Twenty-eight proteins were enriched in three out of three trials. Out of those 28 proteins, three were selected to be further investigated for their significance to sickle cell anemia. With the molecular mechanism of HDAC1 activity in SCD known, targeted drug development will become possible
We Are All Storykeepers: An Exploration Of Intergenerational Trauma, Family Narratives, And Culture
No full textPrevious research has found that trauma exposure is prevalent internationally and may have profound effects. Additionally, the symptoms of trauma exposure can be passed from one generation to the next, potentially through the mechanisms of parenting behaviors and family narratives. The current study drew from these areas of research to examine intergenerational trauma through an attachment-based, family narrative, cultural lens based in qualitative methodology. Ten parents of very young children whose families experienced traumatic experiences took part in interviews. Interviews explored their ancestor’s traumatic experiences, the effects of the traumatic experiences on their parents, their parents’ parenting behaviors, and their experiences parenting their own children. Main findings explored families’ openness or quietness toward talking about traumatic events, impacts of avoiding discussion on processing trauma, generational differences in emotional openness, the effects of avoiding discussion on parenting behaviors, desires to “break cycles” of intergenerational trauma, cultural norms surrounding the discussion of trauma, and the effect of trauma on interrupting cultural engagement. Implications and future directions are outlined
Vision-Language Models For Future Image Caption Prediction: Methods, Applications, And Adversarial Vulnerabilities
No full textImage captioning has traditionally focused on generating descriptions for individual static images. However, predicting future events from visual information is a fundamental challenge in this domain. While existing methods primarily describe current visual content, the ability to anticipate and generate captions for future events remains largely unexplored. We propose a novel approach for future caption prediction by leveraging the capabilities of Vision Transformer (ViT), Generative Pre-trained Transformer 2 (GPT-2), and Text-to-Text Transfer Transformer Model (T5) architectures.
Our method includes two complementary strategies: a two-stage pipeline where ViT-GPT2 generates captions for current images and T5 analyzes these captions to predict future captions, and a single-stage model in which ViT-GPT2 is trained to generate both current and future captions simultaneously from a single image input. We evaluate our approach on a custom Cooking Dataset and compare its performance with baseline approaches. The results demonstrate that our model outperforms baselines in standard caption-generation metrics while offering more contextually rich and anticipatory insights.
Additionally, we examine the vulnerabilities of the proposed multi-stage framework to adversarial perturbations, showing that disruptions in early stages can propagate and degrade downstream performance. We evaluate three attack strategies—FGSM, prompt-based manipulations, and TextFooler—highlighting the cascading effects of adversarial noise and the need for robust design to ensure reliability in real-world deployment
On Compressing Deep Neural Networks
No full textThough deep neural networks achieve great accuracy in visual recognition tasks, they contain millions of weights and thus require a large space to be stored. In this dissertation, we focus on compressing different types of deep neural networks in different situations. First, we present a novel deep compression method, Octave Deep Compression (ODC), to compress Octave Convolutional Networks with in-parallel pruning-quantization on different frequencies. Second, we propose a novel unstructured pruning pipeline, Attention-based Simultaneous sparse structure and Weight Learning (ASWL), where an efficient algorithm is proposed to calculate the pruning ratios layer-wisely from attentions, and both weights for the dense network and the sparse network are tracked so that the pruned structure is simultaneously learned from randomly initialized weights. Third, we focus on compressing and accelerating deep GCN models with residual connections using structured pruning by presenting AgileGCN. Specifically, in each residual structure of a deep GCN, channel sampling and padding are applied to the input and output channels of a convolutional layer, respectively, to significantly reduce its floating point operations (FLOPs) and number of parameters. Fourth, we propose a novel framework, Transferring Lottery Ticket (TLT), to adapt both masks and weights of a pre-trained and pruned network dynamically during the knowledge transfer to downstream tasks. Recent work has shown that pruned networks can also be used as pre-trained models in transfer learning. Finally, we propose MAGNET, a novel modality-agnostic network for 3D medical image segmentation, which is specifically designed to handle real medical situations where multiple modalities/sequences are available during model training, but fewer are available or used at the time of clinical practice
How A 12-Week Multimodal Functional Fitness Intervention For Older Adults Influences Functional Fitness Outcomes While Examining Program Participation Influence
No full textThe purpose of this mixed methods study was to determine how multimodal delivery of a functional fitness exercise program influences functional outcomes in older adults while examining influences on participation adherence. With a rapid decline in functional fitness capacity in older adults, increasing exercise in older adults is evident. Furthermore, understanding exercise delivery modes and their influences on participation may impact accessibility and adherence in older adults. This study focused on answering the following research questions: “How does a 12-week multimodal functional fitness intervention improve functional fitness outcomes in older adults? What multi-level social factors influence participation in both online and in-person delivery of a 12-week functional fitness intervention in older adults?” This study utilizes a mixed methods approach to attain objective functional fitness measures and subjective qualitative data from one-on-one interviews. Quantitative data analysis via a paired sample t-test will determine mean changes in pre- and post-measures across five domains of functional fitness assessment in 38 participants. Thematic analysis of interview transcriptions in 8 participants for the qualitative one-on-one interview was used to gain insight into participants\u27 participation influences. The quantitative findings of this study highlight that multimodal delivery effectively improves upper and lower body strength, cardiorespiratory endurance, and lower extremity flexibility, as measured by the Senior Fitness Test. The qualitative findings give insight into participation adherence barriers and facilitators through participant interviews in older adults, which included several themes related to technological barriers, intrinsic and extrinsic motivators, social support, environmental barriers, and self-efficacy. These insights may be essential in designing multimodal exercise delivery for older adults. Our results illustrate the importance of addressing technological literacy and reducing environmental barriers impacting exercise participation adherence for online access. Our findings highlight promising improvements in functional independence based on criterion-referenced standards for any age, even at low exercise doses