22689 research outputs found
Sort by
USING HETEROGENEOUS INTEGRATION PACKAGING TO DECREASE THE SIZE OF IMPLANTABLE MEDICAL DEVICES
Get PDFThe Bionode has been developed and used for over a decade in the Center for Implantable Devices (CID) Laboratory. Its main goal has been to develop and improve upon a device that can conduct bioelectric signal recording and provide stimulation using wireless power. This device has been the core technology used in the CID lab to study epilepsy, gastric motility, Parkinson's, and various other diseases and has provided further insight into neuromodulation. As the device’s abilities were expanded and enhanced over numerous iterations, the size of the device grew too. For implantable medical devices, a larger implant results in a greater risk for patient irritation and infection, the need for invasive procedures, and increased recovery time. Given that reductions in size, weight, and power can no longer rely on integrated circuit (IC) process scaling alone, new methods are emerging for further miniaturization. Although multi-chip modules (MCMs), similarly known as System-in-Packages (SiPs) as well as wafer level packaging have been around for many years, there has recently been a converge of these technologies to co-package die in chip-scale sizes at high volumes where the dies themselves may come from dissimilar technology nodes. This technique is called heterogeneous integration and is the technology along with other design-related modifications that are applied to decrease the dimensions of the Bionode. A smaller implanted device helps decrease the risk of irritation and infection and allows for chronic implantation for potential experiments. Thus, a smaller Bionode will increase its success rate in chronic implantations and open the possibility to expand research into other animal models
CHARACTERIZING SPATIAL RELATIONSHIPS BETWEEN DISASTER EXPOSURE, SOCIAL VULNERABILITY AND SELECT CHRONIC HEALTH OUTCOMES IN U.S. COMMUNITIES IN PREPARATION FOR CLIMATE CHANGE IMPACTS
No full textObjective: The effects of climate change differ by geography, and due to social and environmental drivers, communities across the United States vary in their preparedness for these hazards. This research explores social vulnerability, frequency of disasters, and chronic disease burden in the United States with the goal of identifying communities in need of additional support adapting to the impacts of climate change.
Methods: This dissertation contains three manuscripts that use tools from the field of spatial statistics and analysis to explore the spatial relationships between social vulnerability, disaster frequency and select chronic health outcomes in United States communities at the county scale. The first and second manuscripts characterize the spatial and statistical relationships of social vulnerability, disaster exposure, and three select chronic health condition prevalence variables, in the contiguous United States and identify multivariate spatial clusters of counties as well as specific counties that are extremely high in multiple or all variables. The third manuscript is a white paper that demonstrates a method for how the findings, methods, and data from this research can be used for practical purposes to isolate specific drivers of vulnerability and develop strategic targeted mitigation and resilience strategies in climate adaptation planning through three example counties.
Results: Multivariate spatial scan statistical analyses that included all five variables revealed two significant clusters. A large cold cluster (low values across all variables) was identified in the central Northern region of the US. A large hot cluster (high values across all variables) was identified in the southeast region of the U.S. Clusters were mapped and overlaid with counties highlighting those that contained values in the highest quartile across all variables. For three example counties, specific drivers of social and environmental vulnerability pulled from data used in this analysis are characterized and used to develop targeted strategies for building resilience to climate change impacts.
Conclusion: This research emphasizes the importance of considering the geography of multiple sectors in building resilience to mitigate the impacts of climate change and highlights the value of multivariate spatial methods in research and practice that incorporates social and environmental determinants of health
Development of Novel Tools for Highly Multiplexed Proteomics and Analysis of Transcriptional Circuitry
No full textThe decoding of the human genome has unveiled a new frontier for comprehending the intricacies of human life. This wealth of information has ushered in a new era of "omics" technologies, refining our understanding of the dynamics of human genes, proteins, and their expression. In pursuit of this, we introduce two novel methodologies: Barcode-Enabled Digital Western, for high-throughput analysis of protein expression in single cells and Proteome-Wide Association Studies, for the analysis of transcription factor binding preferences in the context of complex disease.
The first method, Barcode-Enabled Digital Western (BEND-Western), leverages barcoded antibodies as a readout for intracellular protein expression in single cells. In this method, cells are first lysed in the presence of NHS-magnetic beads to capture cellular proteins. The immobilized proteins are then treated with a cocktail of barcoded antibodies, washed thoroughly to remove non-specific interactions, and then the resulting DNA-barcodes are recovered and analyzed using qPCR with barcode-specific primers or through deep sequencing. This method addresses longstanding challenges in cell biology, surpassing previous limitations in multiplexing capabilities and surface protein focus. Its application in studying heterogeneous cell systems can provide valuable insights into different cell types and their functional states. Experiments detailing the benchmarking of BEND-Western from bulk-cell to single cell level will be described, using model systems such as the cell cycle, organoids, and cardiomyocyte differentiation. BEND-Western also has the potential to be coupled with single-cell RNA-seq, and experiments demonstrating preliminary studies incorporating this analysis into the bulk-cell BEND-Western pipeline will also be detailed.
The second method, Proteome-Wide Association Studies (PWAS), employs transcription factor protein microarrays (TF arrays) to investigate the differential binding of TF proteins to Single Nucleotide Polymorphisms (SNPs) associated with complex diseases through Genome-Wide Association Studies (GWAS). Many SNPs identified through GWAS are within non-coding regions of the genome, and thus it is often challenging to decipher functional variants from the numerous statistically significant variants associated with a particular disease. It has been hypothesized that these SNPs can impact phenotype through altering transcriptional regulation. By identifying TFs that bind differentially with GWAS SNPs, we can begin to provide insight into potential molecular mechanisms underlying different complex diseases. PWAS studies across five different complex diseases will be detailed, including Alzheimer's Disease, Age-related Macular Degeneration, long COVID, Schizophrenia, and Type II Diabetes. Using datasets across complex diseases, we can also gain insight into more broad behaviors of transcription factors in complex diseases, including behaviors of TFs relative to their consensus motifs, TF behavior by family or SNP context, as well as the preferences of TFs across SNPs in a complex disease
OPTIMIZED CLUSTERING: RESOLVING FLUCTUATING STATES OF MATTER
Get PDFNoisy 2D magnetic maze domain state images gathered via holography may include significant low frequency noise which blurs the boundary of magnetic regions in the sample of interest. Under appropriate conditions, the magnetic domains may additionally be evolving between distinct states of matter and a graphical model of state-state transitions may be created.
In this work, we propose a clustering algorithm that optimizes for increased contrast between magnetic regions, identifies outlier images that represent transitions between distinct states of matter, and uses convex combinations of images to reduce noise. We generate synthetic data designed to mimic relevant properties of holography samples of interest and analyze our results generated from a known underlying graph of state-state transitions.
Our method involves first creating a given set of clusters and tuning the convex combination of images within each cluster, and then using the evaluation of this particular clustering and convex combination in order to tune an affinity matrix and improve on further iterations of clustering. Images with very low weights in their convex combinations would be deemed to be outliers, representing transitions between states.
In our results the affinity kernel used for clustering did not well distinguish between distinct clusters, resulting in clustering with an adjusted rand score of 0.2179 when including no transitory frames and 0.2322 when including all images. The convex combinations of images within clusters did significantly improve the custom metric, but the convex combination parameters magnitudes’ were not indicative of state transitions
TEACHER TURNOVER IN CHARTER MANAGEMENT ORGANIZATION SCHOOLS
Get PDFTeacher Turnover in Charter Management Organization (CMO) schools has emerged as a significant challenge, with adverse effects on the financial, academic, and social aspects of schools. This investigates the contributing factors influencing teacher turnover within CMO schools. Employing Bronfenbrenner’s Ecological Systems Theory (EST) as the theoretical framework, the study positions the teacher at the core, examining the factors within influencing sphere systems (microsystem, mesosystem, exosystem, macrosystem, and chronosystem).
Using a mixed-methods approach to collect data, and the Pillar Integration Process to analyze the mixed results, this study addresses six research questions:
1. What are charter school teachers’ perceptions of trust, relationships, and support?
2. How has the COVID-19 pandemic influenced the perceptions of support and trust within a school building among teachers?
3. What are the implications of teacher turnover on teachers’ relationships with teachers and students?
4. What are the implications of school leadership on teacher turnover?
5. How does teacher identity influence teacher turnover?
6. How do teachers’ experiences of trust, relationships, and support influence self-efficacy?
The analysis illuminated four key takeaways.
1. Strong teacher relationships foster trust, influencing turnover.
2. Teacher perceptions of support from leadership influence motivation and trust, with negative perceptions contributing to turnover.
3. Teacher identity plays a significant role in influencing relationships and trust.
4. Perceptions of self-efficacy are shaped by peer and leadership support.
While some limitations existed, such as the sample size and other external factors, this research communicates the importance of addressing teacher turnover in CMO charter schools. By examining factors such as relationships, trust, support, identity, and efficacy, schools can cultivate environments conducive to decreased turnover and increased student success
Statistical Learning of Interaction Kernels in Dynamical Systems
Get PDFWe examine the topic of using data to discover the underlying governing equations of a
dynamical system. With the rise of machine learning and the abundance of data, datadriven discovery plays an important role in multiple fields such as epidemiology, finance
and biology.We first look at a prior approach to data-driven discovery of parameters
in a dynamical system. Then, we introduce systems of interacting agents that are
governed by unknown interaction laws. A statistical learning theory is introduced on
how to infer these interaction laws and computational experiments are performed to
demonstrate the applicability of this approach for inference of the interaction kernel
and trajectory prediction
ASSESSING TWO INTERDEPENDENT RISK FACTORS FOR BREAST CANCER: OBESITY AND DYSBIOSIS
No full textBackground: Breast cancer is the most diagnosed cancer in women and the second-highest cause of cancer-related mortality. A combination of non-modifiable factors such as genetics, family history, age, gender as well as modifiable factors including environmental exposures, obesity and microbiota influence the incidence and progression of breast cancer. Our primary goal was to investigate the influence of two important modifiable risk factors, obesity and microbiota dysbiosis, on breast cancer and delineate the underlying molecular mechanisms.
Methods: We used RT-PCR, western-blotting, and immunofluorescence to test the stemness and signature gene expression levels. The effects of obesity and microbial dysbiosis were also evaluated in vivo in mice. The tumor-dissociated cells from these mice were, thereafter, used for various functional assays for further corroboration.
Result: Our results demonstrated an enrichment of pre-adipocytes in inflammatory breast cancer (IBC), showing a potential link between obesity and a predominant class of breast cancer. We showed that obesity led to increased proliferation and stemness of IBC cells in vivo. Our data strongly implicated the involvement of lipocalin2 (LCN2) as an adipocytokine in IBC progression. Additionally, we found a notable rise in cancer cell proliferation, migration and stemness upon microbial dysbiosis both in vitro and in vivo.
Conclusion: Our collated evidences showed that obesity contributes to the development of IBC via an adipocytokine LCN2, which can, therefore, serve as an attractive target for chemoprevention. Our observations indicated that Fusobacterium nucleatum was overabundant in breast tumors, which enhanced neoplastic transformations, thus strengthening the role of microbial dysbiosis in breast cancer development. Our study, thus, identified the intimate interplay of these emerging risk factors in breast carcinogenesis and our next aim involves further research in these areas for better therapeutic and disease management strategies
ARTIFICIAL INTELLIGENCE-BASED DETECTION AND PREDICTION OF PANCREAS CANCER
Get PDFTexture is a quantitative measure used to describe the spatial arrangement of pixel intensities within an image. However, it is highly dependent on gray level scale. To improve the stability and robustness of texture features, we proposed a normalization method to reduce or eliminate the dependence of texture features on gray levels. Two open-source datasets were used to validate the feasibility of the normalization. The classification results based on the invariant texture features achieved an accuracy improvement of 14.7% and 3.3%, respectively, comparing to using the original texture features. By developing the invariant texture features, an image pattern will yield the same texture feature values independent of the number of quantization gray levels.
Pancreatic cancer is considered as one of the most lethal forms of cancer since it is often diagnosed at an advanced stage. To improve the detection of early pancreatic ductal adenocarcinoma (PDAC) and high-grade dysplasia (HGD) in high-risk individuals (HRI) with a familial or genetic predisposition, we proposed three different pancreatic cancer prediction models, using clinical features alone, invariant texture features alone and a combination of the clinical and texture features, to investigate the performance discrepancy in predicting the diagnosis of PDAC and HGD. Compared to the diagnostic accuracy of clinical factors for predicting PDAC/HGD (50%), EUS texture features had a higher accuracy (78%), and when clinical data and EUS texture features were combined, accuracy of the model to predict progression to PDAC/HGD was 96%. This result demonstrated that the fusion of multimodality data, such as the combination of clinical and imaging models, is more accurate than imaging models, which are more accurate than clinical based models.
Currently, most deep learning based pancreatic cancer prediction models are limited to making predictions at one time point, we piloted the vision transformer-based model that encodes time and allows prediction of future risk of PDAC and HGD. The preliminary result achieved 76.4% accuracy, 82.3% sensitivity, and 62.7% specificity, indicating a promising future work to modify the model to predict the probability of a patient being diagnosed with PDAC and HGD within a desired time interval
MOTIVATIONAL ROOTS OF ACADEMIC STRESS: USING SELF-DETERMINATION THEORY TO EXAMINE PERCEPTIONS OF ACADEMIC STRESS AND HOW SCHOOLS CAN HELP
Get PDFStudents at high achieving schools and cross-cultural students experience elevated levels of stress that contribute to maladaptive physical and mental symptoms. The purpose of this study was to examine student, parent, and educator perceptions of cross-cultural students’ experiences with academic stress, school supports for academic stress, and psychological needs fulfillment. The underlying motivational roots of academic stress were analyzed through an exploration of the connections between academic stress and student psychological needs fulfillment, as outlined by self-determination theory. In this mixed methods study at an elite, international school, quantitative data was collected from educator and parent surveys. Qualitative data was collected from open-ended questions on parent and educator surveys, semi-structured focus groups with students over the age of 18, and semi-structured interviews with students in grades 9-12. Topics of questions and discussions included academic stress, school supports, and psychological needs fulfilment. Descriptive statistics and differences in means were calculated for quantitative data. Qualitative data were analyzed using reflexive thematic analysis. Quantitative data and qualitative data were then merged for areas of convergence and divergence in a mixed methods analysis. Findings revealed students experienced academic stress caused largely by expectations, competition, and workload. In response to stress, students felt exhausted, felt overwhelmed, and developed coping mechanisms. Experiences of academic stress and school supports also appeared to be connected to the degree to which students’ psychological needs were fulfilled. Belonging needs were supported by care and community, autonomy was supported by choice and flexibility, and competence was supported by academic success and encouragement. These findings suggest that high achieving schools with cross cultural populations who are interested in improving student experiences with academic stress may consider implementing needs supportive practices as school supports for stress
Graphical Analysis of Recurrent Surface Temperature Trends Using the Gromov-Wasserstein Distance Metric
Get PDFGraphical networks serve as powerful models for interpreting complex systems by abstracting complex scenarios with a simple network. This work leverages such networks to model the patterns in surface temperature observed within the Gulf of Mexico. To quantify seasonal dynamics, Voronoi diagrams are used to capture sub-regions that share common characteristics. Graphs of these diagrams can then be analyzed like graphs and the Gromov-Wasserstein distance metric captures in a single metric how different a daily graph is from some standard reference graph. These daily snapshots into daily surface temperature trends can then be aggregated and analyzed for relative changes in structure. This method presents a computationally inexpensive method for analyzing oceanic surface trends for evidence of climate change.
While ocean surface temperature is just once data channel in broader climate change work, the methods of this work to provide insight into trends and lay the foundation for broader work in multimodal spatial dataset analysis