22689 research outputs found
Sort by
PRECISION SYNTHESIS AND OPTICAL PROPERTY OF COMPLEX SILICON−SILICON MOLECULAR AND POLYMERIC FRAMEWORKS
No full textThe structure of crystalline silicon has inspired synthetic chemists to design molecular and polymeric cyclosilanes for well-defined novel silicon materials with tunable optical properties. This dissertation describes synthetic strategies of constructing complex ladder cyclosilanes and polymeric cyclosilanes and the relationships between microstructures and optical properties. A series of cyclosilane homo- and copolymers are synthesized by hydrocoupling polymerization. Their optical properties exhibit connectivity-dependence, which results from different σ-conjugation pathways introduced by distinctive linkages. This is the first demonstration of the main chain control of polysilane optical properties. Besides, we report on the syntheses of two series of ladder cyclohexasilanes from mono- to tricyclic structures, which rank among the most complex oligosilanes yet synthesized. Their microstructure-optical property-relationships are examined by experimental and theoretical study. Anion-induced epimerization on cyclosilanes is assumed to go through a pentavalent intermediate. We use 1,2-disubstitued cyclohexasilane to investigate the epimerization mechanism. These results extend the synthetic scope of molecular and polymeric cyclosilanes, as well as deepen our understanding on electronic and optical properties of Si-Si framework
Salinity variability in the subpolar north Atlantic
No full textSalinity is a fundamental quantity which controls the density of the ocean. Fluctuations in salinity, at high latitudes along with temperature affect the global ocean circulation, especially via the Atlantic meridional overturning circulation (AMOC). Changes in salinity thus have important consequences for global climate. Variability in salinity is, however, not well understood. In this thesis, we aim to study the changes in salinity in the subpolar north Atlantic (SPNA), using a suite of modeling tools. We construct salinity budgets using the ECCO (Estimating the Circulation and Climate of the Ocean) state estimate, pre-industrial simulation of the Community Earth System Model 2 (CESM2) and 100 members of the CESM2-Large Ensemble output. Using ECCO, we find that present day anomalous salinity events are driven by distinct mechanisms, and not all salinity anomaly events are the same. We also find, using the pre-industrial simulation that anomalous salinity events in the western and eastern SPNA are driven by a combination of advective convergence and surface forcing, with a larger role of surface forcing in the western SPNA. Finally, we conclude that a major drop of nearly 2g/kg in salinity is projected to occur in the SPNA by the end of 2100. Anthropogenic signals in the vertical mixing and advective tendencies are expected to emerge in the eastern SPNA during 2020--2030 and in the west by 2040
A COMMUNITY-ENGAGED, EQUITY-CENTERED ADAPTATION OF THE CURE VIOLENCE MODEL IN SOUTHWEST PHILADELPHIA
No full textProblem Statement: Community violence remains a major public health crisis, disproportionately affecting Black communities. Traditional responses have relied on policing and incarceration, often exacerbating systemic inequities. The emergence of Community Violence Interventions (CVIs), such as Cure Violence (CV), offers a community-centered alternative that applies public health strategies to disrupt cycles of violence. However, evaluations show mixed effectiveness, raising questions about implementation fidelity, stakeholder engagement, and local adaptation. This dissertation addresses these questions by integrating Implementation Science and community-based participatory approaches to examine how CV, a widely implemented CVI, can be optimized for greater impact.
Methods: The first paper is a scoping review identifying implementation determinants, strategies, and outcomes associated with CV programs, culminating in a Implementation Research Logic Model (IRLM). The second paper details a stakeholder-engaged adaptation process using concept mapping to tailor CV implementation. The third paper, written as a practical implementation report, evaluates the pilot implementation using the RE-AIM framework to assess program reach, fidelity, acceptability, and early effectiveness outcomes.
Results: Findings highlight factors shaping CV implementation and effectiveness. The scoping review identifies knowledge gaps and proposes an adapted IRLM framework. The adaptation study demonstrates a systematic approach to developing strategies in alignment with community priorities, fostering greater engagement and perceived effectiveness. The pilot evaluation incorporates the findings from paper 2 to reveal that while the adapted CV program achieved high levels of community acceptability and engagement, challenges persisted in sustaining fidelity, particularly in violence interruption activities. Participants reported increased access to resources and motivation for change, yet long-term behavior modification was not explored in this study and remains an area for further exploration.
Conclusion: This dissertation underscores the value of integrating IS principles and community-based participatory research methodologies to optimize CV programs. Findings suggest that structured adaptation, ongoing stakeholder engagement, and data-driven decision-making can improve implementation fidelity and have implications for intervention effectiveness. Future research should focus on comparative implementation studies, long-term evaluations of CVIs (including CV), and expanding participatory frameworks to refine community-centered violence prevention strategies. These insights contribute to advancing the field of CVI research and informing policy and practice for sustainable, community-driven violence prevention
OPTIMIZING DESIGN ENGINEERING PROCESSES IN DATA CENTER PRODUCTS: IMPLEMENTATION OF STANDARDIZED WORKFLOWS AND DESIGN AUTOMATION AT TATE INC.
Full text linkEfficient mechanical design processes are vital for the rapidly evolving data
center industry, where precision, scalability, and reliability significantly impact
overall operational performance and competitiveness. Despite advancements,
inefficiencies persist due to inconsistent procedures, limited standardization, and
inadequate use of automation, resulting in prolonged project timelines, increased
errors, and higher operational costs. This essay addresses these challenges through
the implementation of standardized workflows, parametric CAD modeling, and
automated design tools developed for structural Hot Aisle Containment (HAC)
products at Tate Inc.
An initial analysis of Tate’s existing HAC design workflows identified critical
bottlenecks, particularly in manual CAD operations and inconsistent
documentation practices. To resolve these issues, a comprehensive standardized
workflow guide was created, clearly defining procedural steps and integrating
critical checkpoints to enhance communication and reduce design iterations.
Additionally, parametric models were developed in Onshape alongside two
customized automation tools using FeatureScript: "Set Name," for automated part
naming, and "Tab and Slot," for consistent, error-free weld alignment features.
Implementation of these enhancements led to significant qualitative and early
quantitative improvements. Notably, the new standard workflow guide
substantially accelerated the onboarding of new engineers, reducing the
integration period from several weeks to just two weeks. Furthermore, initial data
indicates an 88% reduction in the time required for initial CAD modeling and
submittal drawings, decreasing from approximately 16.5 days to only two days.
Feedback from engineers and stakeholders across manufacturing, procurement,
and project management departments confirmed substantial improvements in
consistency, reliability, and overall clarity of the engineering process.
In conclusion, this essay demonstrates how structured workflow documentation,
creation of parametric CAD models and targeted automation can greatly enhance
mechanical design practices within the data center industry, providing a scalable
and adaptable foundation for future technological integration and innovation at
Tate Inc
AN ASSESSMENT OF HYPERDIMENSIONAL COMPUTING FOR SLEEP STAGE CLASSIFICATION
Full text linkThis thesis investigates whether a selection of hyperdimensional computing (HDC) algorithms is suitable for automatic sleep stage classification using cardio-respiratory input signals to determine their applicability for resource-constrained devices. Three HDC algorithms were evaluated using a single encoder and a pre-trained one-dimensional convolutional neural network feature extractor on four datasets with a total of 1,000 nights of sleep. Performance was reported for four sleep stage classes (wake, rapid eye movement, light sleep, and deep sleep) and was compared with five machine learning baseline algorithms using different combinations of heart rate input signal modalities, electrocardiogram and photoplethysmography, along with abdominal and thoracic respiratory signals.
Performance was compared across training dataset size, where the HDC algorithms did not improve performance, with an average Cohen's kappa ranging from 0.321 to 0.361 across the regimes evaluated. Performance across different signal modalities was also assessed, where a slight improvement in average performance was observed for the HDC algorithms, with an increase in kappa from an average of 0.314 to 0.356, though a greater improvement was observed in the machine learning baselines, with an increase in kappa from an average of 0.336 to 0.480. Generally, the HDC algorithms performed worse than the machine learning baselines in all experiments, and all algorithms performed well below the current state-of-the-art cardio-respiratory model reported in the literature and the minimum threshold for clinical use. Improvements in the HDC algorithms or training methods used in this thesis are necessary for use in automatic cardio-respiratory sleep stage classification.
Future research should consider more complex architectures, such as ones that integrate temporal information within the data, alternative encoding methods, and additional input modalities to improve performance
USING MACHINE LEARNING AND AGENT-BASED MODELS TO INTERPRET AND FORECAST CELL BEHAVIOR IN THE PANCREATIC DUCTAL ADENOCARCINOMA MICROENVIRONMENT
No full textThe single-cell data revolution has sent waves through the fields of tumor immunology and computational biology. New, large data types require significant post-hoc computational interpretation and analysis. The necessary data analysis software is actively evolving along with single-cell and spatial sequencing technologies, and open questions remain as to the optimal algorithmic methods to interpret omics data, especially as it pertains to learning and modeling cell states and emergent cell behaviors. This thesis expands single-cell data science research in two directions: Machine learning to find gene patterns in single-cell data and tease apart the cell states implicated in tumor progression, and agent-based models to forecast cell interactions and behavior from data-defined initial states and behavior hypotheses in order to develop interception strategies. First, non-negative matrix factorization has been widely applied to infer transitions in cellular phenotypes from single-cell data but often requires advanced guidelines to enable widespread application of non-negative matrix factorization and demonstrate its ability to infer transitions in neoplastic cell phenotypes in pancreatic ductal adenocarcinoma. Once discovered, we extended the information about these inferred cellular regulatory networks in pancreatic cancer to use agent-based models to understand how cells in the microenvironment support a series of phenotype transitions that comprise tumor progression and metastasis. I learned that bringing together data-driven and mathematical modeling is a powerful tool in the translational immunology data analysis pipeline because it lets you systematically visualize, predict, and encode emergent cell behaviors seen in omics data. The approaches I present here allowed me to build models to forecast and intercept tumorigenesis and promote successful adaptive effector antitumor immune responses in the pancreatic tumor microenvironment
Longitudinal Relationship between Clofazimine Exposure, Skin Pigment Changes, and Plasma Clofazimine Levels Among Adults Treated for MDR-TB in India
Full text linkThis study investigated the longitudinal relationship between clofazimine therapy, skin pigmentation changes, and plasma clofazimine levels among adults treated for multidrug-resistant tuberculosis (MDR-TB) in India. 697 skin images from a cohort of 91 patients were analyzed. The images were rated using a Munsell Soil-Color chart 10YR hue scale. For each image, a combination of value and chroma was recorded, which was assigned ITA, a, b, and L values as measures of skin pigmentation. Baseline skin tones were defined as skin without the effect of clofazimine, determined by pre- or post-treatment images. The change of each patient’s skin tone was arithmetically and geometrically normalized based on their individual baseline ITA, a, b, and L values to capture within-individual variability over time. It was observed that ITA and the arithmetic normalization method best captured skin tone changes over time. The average time required for peak change in skin pigmentation was between 16 and 18 months after treatment initiation. In contrast, plasma clofazimine trough levels peaked earlier, around 8 months. Stratified analysis revealed no significant sex- or underweight-related differences in time to peak skin pigmentation. However, diabetic patients exhibited significantly earlier skin tone changes. Correlations between skin tone measures and plasma clofazimine levels were consistently weak (R2 < 0.03), suggesting that skin pigmentation is a poor surrogate for plasma clofazimine exposure. Slightly stronger, yet still modest, correlations were observed in diabetic individuals. These findings emphasize that the standard practice of direct blood sampling and plasma drug level measurement remains essential for accurately assessing clofazimine exposure. Overall, this study provides an understanding of how clofazimine affects skin pigmentation and shows that there are complexities to predict plasma clofazimine drug levels by measuring skin tone changes
Metabolic Optimization and Quantification of Ammonia Production in Rhodopseudomonas palustris
No full textAmmonia is an essential compound used in fertilizers and various industrial processes, yet its conventional production remains energy-intensive and environmentally taxing. In this study, Rhodopseudomonas palustris was engineered to enhance biological ammonia synthesis through targeted gene overexpression and transporter knockout strategies. Among all tested conditions, the strain carrying the plasmid pWDS- 2-pBBRR1MCS-2-nifA in the NifA⁎ background produced the highest ammonia levels. This condition achieved an average ammonia concentration of 0.79 mM, with one replicate reaching a peak of 2.34 mM on Day 12, the highest observed value across all experiments. To further enhance ammonia accumulation, the amtB1 gene, responsible for ammonium uptake, was knocked out. The amtB1 knockout in the NifA⁎ background led to a maximum ammonia level of 0.38 mM on Day 14 and retained approximately 25% more ammonia compared to the control, supporting the role of ammonium transport limitation in improving extracellular ammonia concentration. A combined strategy was also evaluated using the NifA⁎ amtB1 knockout strain transformed with nifA and nifA* plasmids. The strain carrying pWDS-2- pBBRR1MCS-2-nifA demonstrated slightly higher ammonia levels than the counterpart expressing nifA* , with an average of 0.26 mM on Day 21. To validate ammonia detection, electrochemical impedance spectroscopy (EIS) was employed. The system displayed a strong linear correlation (R² = 0.9917) between NH4Cl concentration and impedance, confirming its sensitivity and reliability for real-time ammonia monitoring. These findings highlight the pWDS-2-pBBRR1MCS-2-nifA plasmid in the NifA⁎ strain as the most effective approach for maximizing ammonia output. The combination of regulatory overexpression, transporter disruption, and sensitive detection provides a promising platform for future sustainable ammonia bioproduction
SPHERES OF SAFETY: EXPLORING PHYSICAL AND EMOTIONAL WELL-BEING IN SELF AND SOCIETY
No full textDuring my time in the Science Writing program, I grappled with a growing sense of vulnerability—in my work, my community, and my country. These experiences, shaped by personal privilege, in turn shaped my thesis. Only in retrospect did I realize that fear was a central theme in my writing. This work explores how physical and emotional safety influence our identities, decisions, and society. In personal narratives, reported essays, and op-eds, I examine topics such as extreme phobia (individual sphere), workplace bullying (community sphere), and gun violence (societal sphere). Ultimately, my work reflects on stability, resilience, and the urgent need to take care of one another
NOREPINEPHRINE-MEDIATED ASTROCYTE CALCIUM SIGNALING IN FREELY MOVING MICE
No full textAstrocytes are glial cells of the central nervous system (CNS) that play a critical role in maintaining structure and support. While not electrically excitable, astrocytes can signal through dynamic changes in intracellular calcium concentrations which can occur across different spatiotemporal scales. However, under what contexts these astrocytes are engaged as well as the downstream effects of this signaling remains unknown. In this thesis, we use novel imaging approaches to image astrocyte activity in freely-moving mice at microcircuit and subcellular scales. We found distinct responses to different behaviors, and showed that astrocytes are a target of neuromodulatory systems with activity that is tightly correlated with the sleep/wake cycle. We also demonstrated that astrocytes have distinct activity between their cell body and their processes, which may mediate potential downstream effects of astrocyte signaling. This work shows that cortical astrocytes are a potential locus for integrating neuromodulatory and sensory signals, and are well-positioned to modulate neuronal activity and mediate behav-
ioral flexibility. Finally we show that the NMDAR antagonist ketamine dramatically elevates astrocyte calcium in both mice and zebrafish, which may have important implications for understanding the mechanisms underlying ketamine’s antidepressant effects. Together, these
results motivate future endeavors to uncover downstream effects of calcium signaling in astrocytes in different behaviors and to explore their potential as a therapeutic target for disease