22689 research outputs found
Sort by
THE IMPACT OF SCHOOL INTEGRATION SUPPORT ON CHILD PSYCHOSOCIAL WELLBEING IN PEDIATRIC CANCER SURVIVORSHIP
No full textParents of pediatric cancer survivors report challenges with school integration support that may influence their child’s psychosocial wellbeing. Although parents and their experiences of support may be an essential component for survivor psychosocial wellbeing, this relationship has yet to be studied. We conducted an explanatory sequential mixed methods study to answer the research question, how do parent experiences of school integration support affect child psychosocial wellbeing at transition to school following cancer treatment? We first aimed to quantitatively understand how parent experiences of school integration support (parent school integration knowledge and barriers to supportive school integration) related to child psychosocial wellbeing, above and beyond the use of formal education support with consideration to other covariates. We conducted a secondary data analysis and used multiple linear regressions to assess this relationship. The final model (N = 171) demonstrated that use of formal education support (b = -0.352; p = 0.030), Black race (b = 0.615; p = 0.028), and Barriers (b = -0.335, p = <0.001) were significantly associated with child psychosocial wellbeing. We then recruited a new sample of parents to explain our quantitative results and understand the experiences of parents of Black survivors through qualitative individual interviews. We also created a joint display to compare our themes/subthemes across parent experiences and characteristics. In the final sample (N =14) two themes arose from the data: 1) Survivor wellbeing is an interconnected system with three subthemes including, survivor experience, school faculty understanding and communication, and healthcare team communication and; 2) It’s not just support, it’s resources with two subthemes, parent resources and school resources. We found that parents experiences of school integration support are related to their child’s social and emotional health at school transition, both quantitatively and qualitatively. To enhance our psychosocial standards of care for survivors, we must consider the experiences of parents and should measure their experiences of support. Moving forward in practice, validated tools to measure support along with consideration to race and parent access to resources, should be used to connect patients with the appropriate resources to enhance survivor social and emotional health long-term
TARGETING HOST MACROPHAGE HEME-OXYGENASE 1 AND TRIGLYCERIDE PATHWAYS AS HOST DIRECTED THERAPIES AGAINST MYCOBACTERIUM TUBERCULOSIS
Get PDFDisease caused by Mycobacterium tuberculosis (Mtb) remains one of the leading causes of death worldwide from a single infectious agent, killing >1 million people yearly. The 6–9-month antituberculosis drug regimen currently available to treat drug-susceptible tuberculosis (TB) poses a significant challenge, as improper medical adherence contributes to the development of multidrug-resistant (MDR) strains. Host-directed therapy (HDT) offers the opportunity to enhance host defenses against Mtb infection without contributing to antibiotic resistance. Here we investigate the utility of the following HDTs that target macrophage control of Mtb infection: a heme oxygenase-1 (HO-1) inhibitor (SnMP) to enhance macrophage nitric oxide production; and a diacylglycerol acyltransferase 1 (DGAT1) inhibitor and a protein S6 Kinase (p70S6K) inhibitor to reduce macrophage triglyceride (TAG) biosynthesis. Using various murine models of TB disease, we test the therapeutic efficacy of these HDTs as adjunctive agents in combination with TB antibiotic regimens using microbiological, immunological, and histopathological analyses.
We demonstrate in an ex vivo model that inhibition of DGAT1 and p70S6K enzymes consistently reduced intracellular mycobacterial burden and lipid droplet accumulation. We then tested the adjunctive activity of DGAT1 and p70S6K inhibitors in combination with isoniazid in a murine model of chronic TB. We found that, although these HDT agents significantly reduced TAG content in murine lung macrophages, they did not enhance the bactericidal activity of isoniazid or reverse TB-induced lung inflammation.
When testing adjunctive treatment of the HO-1 inhibitor, SnMP, in a mouse model of chronic TB, we find that SnMP significantly enhanced the bactericidal activity of an MDR-TB regimen, modulated the expression of pro-inflammatory cytokines and macrophage-associated genes, and did not exacerbate lung inflammation. Next, we tested the adjunctive sterilizing activity of SnMP in a murine model of microbiological relapse and found that although SnMP 10 mg/kg continued to enhance the bactericidal activity of the MDR-TB regimen compared to the MDR regimen alone after 6 weeks, it did not alter relapse rates.
Together these studies demonstrate the utility and limitations of targeting the TAG biosynthesis and HO-1 pathways as adjunctive HDT for TB disease
Impact Evaluation of Progress Learning in the Douglas County School System
Get PDFThe current study was a retrospective mixed-methods quasi-experimental design (QED) study to determine the effects of Progress Learning on Grades 6-8 mathematics and ELA achievement by comparison growth on the Georgia Milestones Mathematics and ELA assessments of students who received Progress Learning services, in relation to students that did not receive Progress Learning. Supplementary analyses examining the associations between Progress Learning usage metrics and achievement gains are also performed in this study.
The results of the main impact analyses showed a positive and statistically significant impact of Progress Learning on student mathematics achievement, with treatment students outgaining comparison students by more than 4 points. The results of the main ELA impact analysis showed a directionally positive, though not statistically significant, impact on ELA achievement, with treatment students outgaining comparison students by more than 3 points. Effect sizes of these analyses ranged between .06 to .09 SDs, indicating small, though practically meaningful, program impacts of Progress Learning on student achievement, especially in mathematics.
Usage analyses showed significant positive associations between student-level Progress Learning usage metrics and achievement gains. Correlations between average Progress Learning activity scores and achievement gains were of particular note, with observed correlations of magnitude above .4 in ELA and above .6 in mathematics. This gives preliminary evidence supporting modest to moderate predictive validity of Progress Learning activity scores in relation to Georgia Milestones scores. These associations remained significant and positive when controlling for prior achievement and demographics, using HLMs similar to those used in the main impact analyses
MODELING GALVANOTAXIS: PHYSICAL LIMITS OF SIGNAL SENSING & MORPHOLOGY EFFECTS
Get PDFGalvanotaxis -- the migration of cells towards an electric field -- is a natural phenomenon prevalent during many biological processes such as wound healing and embryogenesis. While there has been a recent resurgence in experiments for galvanotaxis, a theoretical understanding is lacking. In this dissertation, I develop models to recapitulate eukaryotic cell migratory behaviors during galvanotaxis, along with developing models that predict cell sensing capabilities with the hope to recreate, optimize, and design new experiments.
In the literature, there is no consensus on how cells sense electric fields. However, experimental evidence suggests that cells sense electric fields via molecules on the cell's surface. These molecules redistribute via electrophoresis and electroosmosis, though the sensing species has not yet been conclusively identified. I develop a model that links sensor redistribution and galvanotaxis for round cells using maximum likelihood estimation. My model predicts how accurately a cell will follow a field. This accuracy is determined by the directionality, which is calculated in my model as a function of the sensor redistribution.
My model also shows how sensor fluctuations can limit galvanotaxis, providing important constraints on sensor properties, and allowing for new tests to determine the specific molecules underlying galvanotaxis. I then extend my model to elliptical cells to account for cell shape and orientation. The model reveals that cells possess more information about the electric field direction along their long axis but may have greater variability in their measurement along the long axis depending on the polarization of their sensors.
Zooming out from the molecular details, I develop a model that predicts the migration patterns of a keratocyte cell, both with and without an electric field present. Keratocytes have been known to oscillate and make turns when there are electric fields present, thus I focus on coupling cell shape and velocity in my model to describe cell migration. My model reproduces cell crawling patterns like persistent motion, oscillatory motion about a field, and circular crawling, and quantitatively recapitulates the cell’s directionality after an electric field is turned on. I make predictions about how cell shape, speed, and stiffness affect cell motility
SYNTHESIS: NAVIGATING SCIENCE AND SELF WITH PLAIN LANGUAGE
No full textWriting in any language is an act of translation. Putting thoughts to paper is an imperfect act, made more difficult when the subject matter is of a technical nature. What results is only an imitation, a best-effort translation, of the writer’s original intentions. The pieces that comprise this work attempt to make topics for a niche audience — such as open access publishing or Great Lakes weather phenomena — universal, paying special attention to accessible and engaging language and structure. In the process of translating scientific concepts for different audiences, this work also showcases the realization of science writing as a lens for personal experience and identity
MODULATION SPECTRUM OF SPEECH FROM LINEAR PREDICTIVE MODELS: APPLICATIONS IN SPEECH RECOGNITION
No full textHumans communicate by modulating the energy in different critical bands. In the modulation spectrum, the slow modulations carry most speech information, with maximum information concentrated around 4-6 Hz. Prior work has shown that these modulations are paramount in human speech intelligibility and effective machine speech recognition. Capturing these slow modulations requires speech analysis over a long contextual window, typically of a duration above half a second. Such long analysis windows capture these slow modulations with sufficient frequency resolution.
However, most traditional speech processing techniques rely on short-term analysis windows lasting only 10-20 milliseconds. As a result, they ignore the critical information embedded in the temporal dynamics of the critical band energies that make up the modulation spectrum. This thesis delves into various aspects of the modulation spectrum of speech, such as its computation and mechanics, information distribution, use in speech recognition, robustness to noise and reverberation, and data-driven modulation analysis methods.
One promising technique for capturing the slow modulations is Frequency Domain Linear Prediction (FDLP) and its variants. FDLP enables the computation of smoothed envelopes of speech signals in different critical bands over long contextual windows, allowing for the derivation of the modulation spectrum of speech. Moreover, FDLP-spectrograms, which are dual to standard short-term spectrograms, compute energy envelopes over long 1.5-second analysis windows, followed by overlap-add of envelopes from contiguous windows. Analysis of the performance of these FDLP-spectrograms in robust speech recognition tasks with noisy and reverberated speech shows how fixing the gain of the FDLP model can help reduce their variability under different recording environments. Further experiments show that even state-of-the-art speech data-driven representation learning methods trained on almost a hundred thousand hours of speech struggle to generalize to unknown noisy conditions. In these situations, FDLP-spectrograms add notable performance improvements and, in unison with state-of-the-art data-driven methods, are universally good at representation learning under different environmental noises.
To investigate the information content of the modulation spectrum of speech further, this thesis explores data-driven and information-theoretic methods. For instance, machine learning algorithms can be trained to learn a weighting function on the magnitude modulation spectrum of speech, achieving the best possible speech recognition performance. The weights learned by this algorithm can then be analyzed post-training. Mutual Information (MI) between the modulation spectrum of speech and the phoneme label at the center of the analysis window can also be computed to determine the amount of information encoded in the modulation spectrum about phonemes.
Finally, the thesis probes into self-supervised feature extraction methods using modulation spectrum. Deep neural networks are shown to be effective in learning missing modulations in speech signals from large volumes of unlabeled data trained in a self-supervised fashion. This approach can enable better speech recognition, especially when limited labeled data is available for training speech recognizers. Overall, the thesis comprehensively explores the modulation spectrum of speech and its potential for improving speech recognition in diverse and challenging environments
EXPERIENCE AND EVOLUTION: FINDING MEANING AT THE INTERSECTION OF NATURE AND NURTURE
Get PDFFrom bass guitars and touchscreen games for dogs to bonobo hands and gray mouse lemur reproduction, this thesis spans a breadth of topics and seeks to examine the influences of nature and nurture on our understanding of the world around us. Forms of writing range from personal essays to more straightforward science news stories. Primatology, psychology, and evolutionary biology play a featured role in most works
Critical behavior of local chemical order in a multi-principal element alloy
No full textChemical short-range order (CSRO) in multi-principal element alloys (MPEAs) constitutes a focal point of current scientific discourse in the metallic materials community. Despite its acknowledged importance, it remains challenging to characterize the local chemical heterogeneity, especially in a quantitative manner, to unveil the origin of the chemical instability and its influence on properties of MPEAs. In this thesis, we develop a quantitative methodology that can analyze the degree of CSRO and concurrently depict corresponding atomic configurations. Using an equiatomic CoCrNi MPEA as a model system, we illustrated the evolutions of CSRO and local atomic configurations with annealing temperatures based on hybrid molecular dynamics and Monte Carlo simulations. Our quantitative analysis reveals a power-law divergence of CSRO at a critical temperature, unveiling the critical behavior of chemical instability within the MPEA. This phenomenon arises from the intricate interplay between entropy-dominant disorder and enthalpy-driven order. To further illuminate the nuanced interplay between entropy and enthalpy, we extend our investigations to non-equiatomic CoCrNi MPEAs. This study not only contributes a nuanced understanding of CSRO evolution in complex alloys but also establishes a novel avenue for portraying the intricacies of local chemical order in MPEAs
DEVELOPMENT OF CHIMERIC ANTIGEN RECEPTOR T CELLS FOR TARGETING GENETIC ALTERATIONS IN CANCER
No full textChimeric antigen receptor (CAR) T cells have revolutionized the treatment of B cell malignancies and multiple myeloma. However, the development of CARs for other cancer types, especially solid tumors, has been hampered by a lack of sufficiently cancer-specific antigens that avoid deadly on-target, off-tumor toxicities. The genetic alterations that underpin the development of cancer offer one way to unequivocally distinguish cancer cells from their normal cell counterparts. Here we identify two classes of genetic changes in cancer that can be targeted by novel CAR designs. The first class of targets are genetic deletions that result in the loss of heterozygosity of some genes in cancer cells. We describe a NOT-logic-gated CAR T cell that can target the loss of a human leukocyte antigen (HLA) allele on cancer cells while sparing normal cells with heterozygous HLA alleles both in vitro and in a mouse model. The second class of targets are point mutations in common driver genes, such as the TP53 R175H hotspot mutation, that are processed and presented on the cell surface as mutant-peptide-HLA complexes. We create a receptor that combines the antibody binding domain and co-stimulation of conventional CARs with the multi-subunit T cell receptor to potently target mutant-peptide-HLA molecules despite their low-density on the cancer cell surface, resulting in the eradication of tumors in mice that are resistant to T cells equipped with patient-derived TCRs. These two new CAR strategies demonstrate complementary methods of leveraging genetic changes to specifically redirect T cells against cancer cells and will hopefully form the foundation of clinically translated therapies in the future
GENE REGULATORY NETWORKS GOVERNING MULTIPOTENCY OF ADULT MURINE BONE MARROW STROMAL CELLS
No full textRecent advancements in single-cell RNA sequencing (scRNAseq) have provided substantial insights into the heterogeneity of bone marrow stromal cells (BMSCs). However, the definition of multipotent BMSCs and their regulatory mechanisms remains a subject of contention. In our study, we utilized the StemFinder tool to analyze compiled data from 20 independent scRNAseq datasets, revealing Hmgb2+ BMSCs as the potentially most potent population. However, lacking definitive lineage tracing evidence, this prediction retained some uncertainty.
To identify specific multipotent BMSC populations, we performed single-nuclei multiomics analysis on early trilineage differentiating BMSCs. The analysis uncovered a transcriptomic distinct cluster of multipotent stromal cells (SCs) unlike the Hmgb2+ BMSCs. This intriguing discrepancy prompted a deeper investigation of freshly isolated BMSCs using single-nuclei multiomics, revealing their rapid transition into an transcriptomic distinct expanded BMSC state within 24 hours of isolation. Trajectory analysis suggested Hmgb2+ BMSCs as the origin of these more potent expanded BMSCs, hinting at a possible dedifferentiation process.
By performing the gene regulatory networks analysis on expanding and differentiating BMSC, utilizing both transcriptomic and chromatin accessibility information, we identified candidate regulators of BMSC fate, including PPARG and NFKB-related factors, and highlighted the significance of Hippo pathway related factors AP1 and TEAD1 as key players in transforming BMSCs into multipotent SCs. Our model predicted that the Hippo pathway regulates BMSC multipotency by coordinating downstream TEAD and AP1 factors. We suppressed Hippo pathway activity using TEAD inhibitors and revealed notable enhancement in BMSC differentiation potential. Conversely, when we enhanced the Hippo pathway using LATS inhibitors, we observed opposite results, confirming the pivotal role of the Hippo pathway in maintaining multipotency.
In summary, this study presents valuable multiomics resources regarding bone marrow stromal cells (BMSCs) and delineates the state transitions during BMSC trilineage differentiation and early expansion in vitro. These findings offer crucial insights into BMSC multipotency and highlight the Hippo pathway transcription factor TEAD as a crucial regulator in this process