22689 research outputs found
Sort by
Measurements of the Largest-Scale Polarization and Temperature Evolution of the CMB
Get PDFThis thesis presents the development and recent progress of the CLASS (Cosmology Large Angular Scale Surveyor) telescope, with a particular focus on its 40 GHz survey. CLASS is a four-frequency telescope array located in the Atacama Desert in northern Chile, designed to observe the cosmic microwave background (CMB) polarization at 40, 90, 150, and 220 GHz. The work details the development of the data reduction pipeline, including signal demodulation from the novel variable delay polarization modulator and map-making. It also highlights the characterization and mitigation solutions for systematic effects arising from both the instrument and the environment. The thesis concludes with an analysis of the CMB temperature evolution using measurements from the Atacama Cosmology Telescope, demonstrating the versatility of ground-based CMB experiments in understanding cosmology
Open Mic with LISA: Intermediate-Mass Black Holes and Double White Dwarfs
Get PDFThe Laser Interferometer Space Antenna (LISA) is a space gravitational wave (GW) detector that will operate in the frequency band 0.1–1,000 mHz. As a GW detector, LISA is not a directional instrument. Instead, it can be thought of as a big “microphone” that will be recording a myriad of GW “sounds” at the same time. Among those are the voices of two rich yet scarcely known populations we focus on in this work: intermediate-mass black holes (IMBHs) and Galactic double white dwarfs (DWDs).
First, we show that it is possible to use LISA for indirect detection of IMBHs in the Milky Way if they are orbited by binaries emitting GWs in the LISA band. The GW signal from the binary around an IMBH will undergo a Doppler modulation which may reveal the presence of the IMBH. We find that the globular cluster (GC) Omega Centauri, which is long suspected to host an IMBH, is the most promising candidate for a positive detection with this method.
Second, we investigate a scaled-up version of the three-body systems considered above, in which the supermassive black hole (SMBH) in our Galactic Center is orbited by an IMBH paired with a stellar-mass object. We demonstrate that the induced Doppler modulation can then be used to constrain the presence of IMBHs with masses 1,000–100,000 solar masses at a distance of 0.1–2 milliparsecs from the SMBH.
Third, we focus on the quasimonochromatic nature of the Galactic DWDs: the fact that, for many of them, the GW frequency drift during the LISA observation time ∆f ≲ 1/yr. We revisit the stationary-phase approximation (SPA) widely used in estimating measurement uncertainties in the frequency domain (FD), and we show how the conventional SPA must be modified for consistent results.
Finally, we investigate whether the Galactic DWDs can be used to probe the Milky Way’s large-scale gravitational field. While it is unlikely to measure the respective center-of-mass accelerations with GWs alone, combining GW and electromagnetic (EM) observations can make the measurement possible. We also propose an efficient implementation of the Fisher matrix calculations which makes use of auto-differentiation.
In this thesis we focus on papers [1–4] where the author made major contributions. To a lesser extent, the author also contributed to papers [5–7] over the course of the PhD program
THE EFFECT OF POLYPHARMACY ON LDL LEVEL AMONG ELDERLY DIABETIC PATIENTS IN THE ATHEROSCLEROSIS RISK IN COMMUNITIES (ARIC) STUDY
Get PDFBackground: Type II diabetes patients often have dyslipidemia, which is a risk factor of atherosclerotic cardiovascular disease. In turn, CVD is known as the leading cause of mortality among this population, thus understanding the changes in LDL levels among them is crucial. The elderly diabetic population often requires multiple medications due to common comorbidities leading to poorer physical health, but research is limited on whether taking multiple types of medications affects LDL levels.
Objective: To evaluate the effect of polypharmacy on lipid profiles among elderly participants with type II diabetes in the ARIC Study.
Design: 2332 ARIC participants over 65 years with type II diabetes were included (from 1990-2017). The primary outcome was change in LDL. Change in HDL and triglycerides were secondary outcomes. Polypharmacy was the exposure of interest, defined as the total number of medications a participant took simultaneously. In the primary analysis, a complete case analysis with a GEE model were applied to fit the effect of polypharmacy on the three outcomes. Multiple imputation was used when evaluating the sensitivity of results to the exclusion of incomplete cases.
Results: For a patient with moderate polypharmacy, LDL levels increased by 0.09 SI unit (95% CI, 0.09 to 0.09); with severe polypharmacy, it increased by 0.21 SI unit (95% CI, 0.21 to 0.21). Moderate polypharmacy had a positive but minimal effect (4.81E-03) on HDL levels. A patient’s HDL levels would decrease by 0.02 SI unit (95% CI, -0.02 to -0.02) with severe polypharmacy. Triglyceride levels would decrease by 0.02 SI unit (95% CI, -0.02 to -0.02) under moderate polypharmacy but increase by 0.04 SI unit (95% CI, 0.04 to 0.04) under severe polypharmacy. The results in sensitivity analysis suggested the main effects of polypharmacy were robust to missing values, as the differences between the estimates were not statistically significant.
Conclusion: Polypharmacy influenced the lipid profiles negatively, including increased LDL levels, decreased HDL levels, and increased TG levels (under severe polypharmacy), the changes were statistically significant but may not be clinically meaningful due to the small magnitude
A NOVEL APPROACH TO COMPUTATIONALLY EFFICIENT AND TRANSPARENT MULTIVARIATE LINEAR MIXED-EFFECTS MODELING WITH APPLICATION TO ESTIMATE THE TREATMENT EFFECTS OF A SCHIZOPHRENIA ANTIPSYCHOTIC DRUG
No full textThe linear mixed-effects model (LMM) is a common statistical approach for analyzing repeated measurements over time. It estimates the effects of predictor variables on the mean level and trend of an outcome variable while accounting for heterogeneous trajectories among participants. In the context of multiple outcomes, a multivariate linear mixed-effects model (MLMM) is an appropriate but less commonly used alternative to separate LMM models, potentially improving the efficiency of fixed and random effects estimates by optimally exploiting associations among the multiple outcomes of interests.
This thesis introduces a novel, iterative, and transparent estimating equation algorithm (EEA) for fitting MLMMs to a specific subset of designs with common observation times to estimate time-dependent treatment effects. The method is motivated by a multivariate analysis of three subscales of the Positive and Negative Syndrome Scale (PANSS) in a cohort of schizophrenia patients receiving either risperidone or placebo. The estimating equation estimators of the random effect and residual variance-covariance matrices are similar to those obtained using univariate LMMs on parameters within each outcome. The EEA also demonstrated comparable statistical performance to the Bayesian method in a simulation study. In the case where the effect of treatment on the total PANSS score is the focus, using the EEA to fit a MLMM offered a negligible benefit in efficiency and no enhancement in power of hypothesis testing
School Belonging Among Traditionally Marginalized Secondary Students
No full textStudents from racial and ethnic minority groups often perceive educational environments as unwelcoming and unsupportive, leading to academic underperformance, decreased school belonging, and psychological issues like depression and anxiety. Factors such as stigmatization, racism, stereotype threats, poor academic fit, and microaggressions negatively affect these students’ experiences and weaken their connection to school. I conducted a two-phase study in one high and two middle schools in a small, suburban city in the northeastern United States. In Phase 1, I used the bio-psycho-socio-ecological model and a retrospective pretest-posttest design to analyze students’ (n = 929) perceptions based on secondary survey data. There were statistically significant differences in students’ perceived belonging by school year, race/ethnicity, gender, and grade level. Implications for further research included further examination into the environmental and interpersonal factors that may negatively affect the schooling experiences of marginalized students. In Phase 2, I created the Interconnected Model of School Belonging Among Traditionally Marginalized Students and employed a phenomenological design to understand adult stakeholders’ perceptions of the supports and barriers to cultivating student belonging and how they contribute to student belonging. Semistructured interviews were conducted with four parents and administrators; transcripts were analyzed using thematic analysis. Results indicated that traditionally marginalized students encountered both proximal and distal (e.g., societal racism, discrimination) challenges that influenced their sense of school belonging. School leaders enhanced the school environment through strategic policies, supportive curricula, and by actively engaging with families. Teachers influenced belonging by fostering engaging learning experiences, creating collaborative opportunities, and developing strong home–school partnerships based on trust and empathy. Peers fostered belonging as they formed networks with others who shared similar backgrounds and interests. Implications for practice include adopting inclusive curricula, enhancing teaching quality through targeted professional development, fostering responsive instructional leadership, and implementing equitable disciplinary practices. Research implications highlighted the need for qualitative studies that examine other marginalized students’ perspectives (e.g., non-native English speakers, LGBTQIA+ individuals) on school belonging and studies of the roles of diverse school leaders and the experiences of immigrant families to comprehend broader systemic factors affecting student engagement and belonging
CHARACTERIZATION OF URINE-DERIVED CELLS CULTIVATED IN KIDNEY AND MESENCHYME MEDIA
Get PDFAbstract
Previous studies have shown the utility of urine-derived cell cultures (UDCCs) to study regenerative medicine and kidney diseases. Here, the aim was to culture cancer cells from the urine of renal cell carcinoma (RCC) patients. To accomplish this aim, we optimized culture conditions including media and extracellular matrix to grow different cell types from the urinary system. The effect of donor age was also examined. Specific gene expression was measured to assess the cell types.
UDCC from the urine of donors was generated by centrifuging urine and washing the pellet with a balanced salt solution. Cells were plated in 24-well dishes and expanded to 6-well dishes in kidney or mesenchyme media. At confluence, RNA was extracted, and RT-qPCR was performed to assess the expression of kidney, urothelial, and mesenchyme-specific genes. RCC-specific genes were also measured. Immunocytochemistry was conducted to validate different cell types using antibodies specific to kidney, epithelial, and mesenchymal proteins. Cells were identified by DAPI staining.
The relative gene expression (ΔCt) was determined by subtracting the Ct value of the reference gene (GAPDH) from the Ct values of the specific genes. Relative gene expression was compared to negative (IMR-90 and HEK293 lines) and positive controls (kidney proximal tubule and UM-UC14 bladder cell lines). In the immunocytochemistry experiments, the percentage of cells expressing different proteins was computed by dividing the number of cells stained with antibodies by the total number of cells.
Higher levels of urothelial cells, renal tubular cells, and mesenchymal stem cell-specific gene expression were observed compared to negative controls. Donor age, media, and collagen matrix. did not affect expression. Notably, higher levels of RCC-specific genes were observed in UDCC from RCC patients compared to age-matched controls.
The limitation of our study was that only three UDCCs were analyzed. Nonetheless, our study suggests the utility of UDCC in the clinical care of RCC. Because cancer cells from RCC escaped into the urine, they are potentially metastatic. Consequently, studying UDCC from RCC patients could be helpful in clinical management. Moreover, the clinical care of other cancers of the urinary system might benefit from establishing UDCC
EARLY TIMEPOINT ANALYSIS AND CLASSIFICATION OF CT PERFUSION DATA PREDICTING DSA COLLATERAL LEVELS
Get PDFStroke, a devastating neurological condition, is the world’s leading cause of disability and death. In the challenging environment of stroke management, where every second counts, the value of precise and prompt diagnostics is undeniably apparent. Imagine a medical professional analyzing CT Perfusion scans minutely while working against the clock to determine the best course of treatment for a patient who has just suffered a stroke. The scan’s delicacy, which reveals crucial details about blood flow and tissue viability, serves as crucial pieces in a puzzle that can change a person’s life. The urgent need to provide healthcare professionals with precise, dependable tools to decode these scans quickly and accurately, as well as a keen awareness of such crucial moments, led to the conception of this project.
This project aims to create a solution that not only improves diagnostic accuracy but also acts as a steadfast ally for clinicians through the lens of arti- ficial intelligence and neural networks. By creating an AI-based algorithm to predict DSA scores from CT perfusion scans, this project aims to navigate the complex decision-making that takes place in the crucial first hours following a stroke. By fusing technology and medicine, it envisions a future where timely, well-informed decisions are an essential part of stroke management
A MOSSY FIBER NEURAL DECODER: DECODING TARGETED SACCADES USING LSTM NETWORK
Get PDFSaccades are rapid eye movements that change the fixation center of the eye from one direction to another. Previous research has proved that mossy fibers of the burst-state neurons in the cerebellum contained kinematic-related information about eye movement during the saccade. However, decoding the exact position and velocity of the eye during a saccade relying only on the neural firings from mossy fibers remains a challenge. This paper introduced a novel design of a mossy fiber decoder based on a long-term memory neural network that reproduces the movement trajectory of the eye during a saccade given real-time neural firings, initial position, and eye velocity. The neural decoder steadily predicts eye pupil trajectories of the marmosets' primary, secondary, and returning saccades
THE RESPONSE OF THE PROSTATE TUMOR IMMUNE MICROENVIRONMENT TO NEOADJUVANT HORMONAL THERAPY
Get PDFProstate cancer is a life-threatening disease to men and treatments like chemotherapy and
hormonal therapy are often used. Androgen deprivation therapy is a type of hormonal therapy
to block androgen activity by inhibiting androgen synthesis or AR transcription factor activity.
During our study, androgen deprivation therapy is used and analyzing cell changes in the
prostate tumor microenvironment following treatment is very important to understand the
cellular and molecular association of treatment response
Mechanisms of forest decline and efficacy of prevention strategies in the Santa Fe Mountains, New Mexico
Get PDFIn recent decades, a rapid escalation of atmospheric drought, coinciding with a global temperature increase, has resulted in heightened rates of tree mortality attributed to wildfires and bark beetle infestations, exacerbated by heat and drought stress. In the United States, these effects are magnified by unnaturally dense forests after over a century of fire exclusion. Sustained losses may prompt numerous regions to shift from being carbon sinks to carbon sources, as numerous forests in the southwestern United States transition into shrublands. Fuel removal treatments are conducted across the United States, including in the Santa Fe Mountains of New Mexico, aimed at mitigating these impacts and increasing resilience to avert further losses. These treatments are known to reduce fire severity and enhance ecological resilience in many cases. However, as global warming persists, there remains a lack of understanding regarding the unintended consequences of these projects, especially in arid and semi-arid locations. For instance, in 2022, two separate prescribed fires inadvertently led to the largest wildfire in the history of the state. This study employs a time series analysis and random forest machine learning model to assess recent changes in a study area in the Santa Fe Mountains. The analysis focuses on land surface temperature and vegetative greenness at 385 study points impacted by spruce beetles, wildfires, and fuel treatments, with untreated areas as controls. The machine learning model predicts the severity of the 2022 Hermits Peak/ Calf Canyon Fire if the perimeter had spread closer to the city of Santa Fe. Results indicated some treatments may be facilitating forest loss, but lighter treatments may offer similar fire severity reduction with less ecological cost, highlighting the need for increased understanding to avoid predictions of near-total forest loss in the area by 2100