Environmental and Occupational Health Sciences Institute
Rutgers University Community RepositoryNot a member yet
58345 research outputs found
Sort by
Towards generalist medical artificial intelligence: model, data, knowledge, and beyond
No full textThe field of medical image analysis has been revolutionized by deep learning, yet current artificial intelligence (AI) systems still fall short of human expert capabilities in complex scenarios. This dissertation addresses three fundamental challenges in bridging this gap: developing robust and accurate model architectures for medical imaging, addressing data heterogeneity and learning from limited datasets, and integrating structured domain knowledge into AI systems. To address these challenges, we first advance model architectures through novel attention mechanisms and shape-aware learning frameworks that effectively capture multi-scale features while incorporating anatomical priors, demonstrating superior performance across diverse imaging modalities. We address data challenges through automated augmentation strategies and a universal learning paradigm that enables knowledge transfer across different anatomical structures and imaging protocols, significantly improving the model's ability to handle diverse tasks. For knowledge integration, we develop explainable frameworks that incorporate structured domain expertise into the learning process, enabling interpretable decision-making that aligns with expert reasoning patterns. Comprehensive experiments across multiple imaging modalities and anatomical targets demonstrate the effectiveness of our approaches, consistently outperforming existing methods while providing interpretable results. This work represents significant progress toward more capable AI systems that can better support medical image analysis through improved accuracy, efficiency, and explainability, while establishing a foundation for future developments in universal medical understanding, integrated vision-language systems for automated diagnosis, and holistic analysis through multi-modal data integration.Ph.D.Includes bibliographical reference
Investigating stress-induced changes to the microstructure of motivated behavior
No full textThirty-three million people experience major depressive disorder (MDD) in their lifetime, but only about 30% of patients respond to current treatments. Anhedonia is a common symptom of MDD, which includes a lack of motivation for once rewarding activities. Furthermore, MDD is typically precipitated by chronic stress. Effort-related choice (ERC) is a useful paradigm for measuring motivational deficits in human and animal populations, in which the animal must choose between a high effort/high reward or a low effort/low reward. In this study, mice were subjected to chronic non-discriminatory social defeat stress (CNSDS) and completed an ERC task. The behavioral microstructure, including behavioral syllable frequencies and transition probabilities, was compared between CNSDS and control conditions. The results indicate the control mice had greater syllable frequency and transition probabilities compared to the CNSDS-exposed mice. This implies that the control condition had a more consistent behavioral strategy compared to the chronically stressed mice when deciding between a high effort/high reward and a low effort/low reward. Future directions include comparing the behavioral microstructure of males and females within both conditions.M.S.Includes bibliographical reference
Interactions between underrepresented students and STEM faculty during the COVID-19 pandemic
No full textThe success of underrepresented students in science, technology, engineering, and mathematics (STEM) fields is a topic of concern in higher education. Student-faculty interactions are widely recognized as a support for these students, but the adoption of emergency remote teaching (ERT) during the COVID-19 pandemic changed how these interactions took place. This study used a mixed methods design, analyzing data from the National Survey of Student Engagement (NSSE) to determine how student-faculty interaction frequencies changed nationally during ERT. Additional semi-structured interviews were held with underrepresented STEM students and STEM faculty members at a large research university to provide further context on these overall trends. Key findings from the NSSE data include a decrease in student-faculty interaction frequencies during ERT, while key differences among student populations by gender identity, race, and ethnicity remained. Students recounted frustrations over additional obstacles to talking with faculty during ERT, while recognizing the efforts that some instructors made to support them. Faculty members in turn recounted their difficulties with observing students' reactions to their instructional methods, while appreciating the opportunity to learn new instructional techniques and gain a greater understanding of their students' lives. These and other data collected from this research are examined through a conceptual framework using Tinto's (1975) student retention theory, Rendón's (1994) validation theory, and related constructs.Ph.D.Includes bibliographical reference
“My job is the TV”: an ecological approach to intellectually disabled adults’ media use
No full textThis dissertation is an ecological systems theory approach to the media use of intellectually disabled adults living with family members under guardianship. Within a systems theory framework, this project utilizes access theory, usage or gratification theory, and parental mediation theory to analyze and contextualize how intellectually disabled adults are active within structures including the accessibility of media, guardian regulation, and resource eligibility. I describe ways in which systems impress upon the central intellectually disabled individual and how the individuals simultaneously impress themselves against the systems. Findings from this study address the present gap in research around intellectually disabled adults both in terms of their disproportionate absence within existing academic research and of the propensity to talk around disabled participants instead of including them in the research process.Using an inclusive research methodology developed alongside community insiders, recruited families were engaged over the course of four interviews and a completed media journal. Multiple interviews with intellectually disabled adults, their primary caregivers, and additional household members ensured information triangulation. Data was analyzed along a flexible coding model within an ecological systems framework. Findings reveal numerous reported gratifications associated with media use and simultaneous barriers including inaccessible technology, resisted caregiver moderation, and insufficient available programming. Participants report gratifications gained from entertainment media use including sense of control, tension relief, sense of accomplishment, sense of independence, and relationship maintenance. Media is reportedly a source of connection with family and friends through co-use, formulaic interactions, and empowerment through bottom-up caregiving. Simultaneously, disabled adults and their guardians report conflict stemming from caregiver moderation and access facilitation of media due to inaccessible technologies and safety concerns.
To analyze the stakes of accessible communication media, and the volume of media reportedly used by participants, this dissertation examines how larger structures, like caregiver networks and government institutions impact media use. The quantity, quality, and transportation to programming was found to be correlated with the cultural capital acquired by guardians through networking. Due to the reportedly convoluted nature of institutions for intellectually disabled adults, caregivers form their own information networks to share resources. The state of insufficient programming options for intellectually disabled adults results in not only a higher volume of media use but a reported preference for media use over existing alternatives.
The findings from this dissertation demonstrate the value of media for this population. In this dissertation's conclusion, I articulate future work needed to ensure that the promised potential of media for this population is reached. The reported barriers to media necessitate improved access of content and form in addition to media trainings for both intellectually disabled adults and their caregivers. Finally, the gratifications reported from media use all stemmed from content not created for intellectually disabled adults. While participants creatively utilized it to meet their needs, they should have access to media created for them.Ph.D.Includes bibliographical reference
Carceral Identity and it's influence on political attitudes and behavior
No full textThis study draws on social identity theory to introduce "carceral identity," a crucial but overlooked social identity that forms among those with sustained contact with the criminal justice system. Despite decades of prior research on social identity, to the best of my knowledge this dissertation is the first scholarly work to make this connection. A criminal record, particularly when coupled with incarceration, carries a heavy stigma that negatively impacts one's self-concept. Yet, this stigma can also forge a sense of solidarity among those who share these experiences, driving them to connect with one another to normalize their experiences and even mobilize for change. This study argues that varying levels of attachment to carceral identity significantly influence the development of group consciousness which, in turn, shapes political attitudes and behavior. To explore this, semi-structured interviews uncovered the conditions under which carceral identity forms and evolves into group consciousness. Next, an original survey, administered via snowball sampling and in partnership with YouGov, yielded responses from over 1,000 individuals with incarceration experiences. This unique sample provides evidence that stronger carceral identity and group consciousness lead to generally more progressive attitudes on criminal justice policy and greater activism on criminal justice and other issues but correspond with lower participation in the voting booth. By introducing carceral identity into the discussion, this study offers a fresh take on the inconsistent findings in the literature on the political behavior of those impacted by the criminal justice system.Ph.D.Includes bibliographical reference
Identification and characterization of stress-induced small proteins using translation initiation profiling
No full textHigh-throughput next-generation sequencing technologies have greatly expanded the gene expression data and enabled the discovery of previously unannotated proteins. Small proteins—those 50 amino acids or fewer—are an emerging class of stress response regulators. Escherichia coli encodes over 150 small proteins, most of which lack defined functional information, and their biological roles remain largely unknown. The main goal of this dissertation is to identify and characterize small proteins that are expressed in a condition-specific manner. In this study, I optimized a specialized ribosome profiling method known as Ribo-RET to map the translation initiation sites of different proteins. Utilizing this method, I identified the subset of previously annotated small proteins expressed in response to low magnesium stress in E. coli. I found 17 small proteins with increased translation initiation under magnesium limitation, several transcriptionally upregulated by the PhoQ-PhoP two-component signaling system, crucial for magnesium homeostasis. Next, I systematically performed deletion and overexpression analysis to identify small protein-specific phenotypes, underscoring their physiological significance in low magnesium stress. Most remarkably, I elucidate an unusual connection via a small membrane protein YoaI, between major signaling networks – PhoR-PhoB and EnvZ-OmpR in E. coli. This thesis advances our understanding of small proteins by shedding light on their regulatory roles under magnesium stress, bridging critical knowledge gaps in their function and physiological relevance. These findings lay the groundwork for future research into their evolutionary origins, mechanisms of action, and potential applications in antimicrobial strategies.Ph.D.Includes bibliographical reference
Born to run: motherhood appeals, race, and partisanship in congressional campaigns
No full textThis dissertation examines the value of motherhood appeals for congressional candidates and explores the question as to whether motherhood is a universal beneficial appeal for candidates. I examine how Black, Latina, and white candidates with young children invoke their motherhood while running for U.S. Congressional seats between 2018 and 2022. The overarching question driving this research is: How does motherhood of dependent-aged children intersect with racial and ethnic identities to shape political opportunities for women candidates? Using a mixed-methods approach, I conduct in-depth case studies of six non-incumbent mother congressional candidates who ran between 2018 and 2022, analyzing campaign materials through a typology of motherhood appeals, and I conduct an experimental survey of voter perceptions. Findings reveal that candidates’ motherhood appeals are neither monolithic nor universally received. Candidates strategically adapt motherhood rhetoric, transforming existing frames like "Mama Grizzly" into broader protector narratives while also introducing novel appeals like Integrated Motherhood and Model Minority Mom. These appeals intersect with candidates’ race, ethnicity, and ideology, reflecting distinct strategies to connect maternal identity to political goals. Case studies highlight how external factors—such as campaign competitiveness and opponent dynamics—moderate the use of motherhood appeals. Experimental findings demonstrate that voter perceptions of mother candidates vary significantly depending on appeal type, racial identity, and partisan alignment. Results indicate that the Model Minority Mom appeal is consistently associated with higher voter support, especially Republicans while Maternal Grief appeal generally reduced support, particularly for Black candidates. This research complicates assumptions about candidates’ motherhood appeals and their strategic use, revealing them as tailored appeals aimed to negotiate racial and gendered stereotypes, align with cultural values, and resonate with diverse voter groups.Ph.D.Includes bibliographical reference
Development of genetically encoded sensors for real-time monitoring of GLP-1 dynamics in the paraventricular nucleus of the hypothalamus
No full textGlucagon-like peptide-1 (GLP-1) is a critical neuroendocrine signal involved in glucose homeostasis and appetite regulation. Here, we report the development of genetically encoded fluorescent GLP-1 sensors (GLP-1 RTGR) designed to monitor GLP-1 dynamics in real time within the paraventricular nucleus (PVN) of the hypothalamus in freely behaving mice. By engineering GLP-1 receptors fused to circularly permuted fluorescent proteins, we achieved highly specific and sensitive detection of endogenous GLP-1 release. In vitro, ex vivo, and in vivo validation confirmed that these sensors exhibit robust and selective responses, enabling precise spatiotemporal mapping of GLP-1 fluctuations. Fiber photometry measurements revealed dynamic GLP-1 signaling patterns that correlate with feeding states, increasing postprandially and potentially reflecting satiety signals. Furthermore, preproglucagon knockout models lacking endogenous GLP-1 showed disrupted feeding patterns, underscoring GLP-1’s role in appetite control. Using a moving average crossover analysis, we predicted feeding behavior from GLP-1 fluctuations with high accuracy. Our findings demonstrate that GLP-1 RTGR sensors provide a powerful tool for real-time neuropeptide imaging, offering new insights into the neuroendocrine mechanisms governing energy balance and highlighting potential targets for metabolic disorder therapies.M.S.Includes bibliographical reference
Applications of LC-MS based metabolomics in pharmaceutical science
No full textThis dissertation is to support the Degree of Doctor of Philosophy in Pharmaceutical Science at Rutgers, The State University of New Jersey. It documents my research work conducted under the guidance of Dr. Ah-Ng Kong in the Department of Pharmaceutics. The work presented here is original, except for where suitable references are made to previous work. This dissertation consists of five chapters and one appendix chapter. They are either published manuscripts or intended to be submitted for publication to journals indexed by PubMed Central. Chapter 1 provides the background for my research. It discusses the importance of metabolomic studies in biomedical research. Metabolomics describes the end products of biochemical processes which are greatly influenced by genetic and environmental factors. Metabolic alterations can be linked to potential biochemical reactions/enzymes and their corresponding genes. Thus, these results can be further validated via a multi-omics approach including genomics, transcriptomics, and proteomics. Metabolomic analysis of biological samples exhibits the possibility to determine the mechanism of action of anti-cancer agents, biomarker discovery, and impact of genetic alterations. Specific emphasis is on evolution of metabolomics in cancer research, metabolomics workflow, general understanding of LC-MS based platform for quantitation of metabolites and their biological interpretation.
Chapter 2 discusses the first application of metabolomics in pharmaceutical science and has been published. Herein, we investigated the effect of HDAC inhibitor (HDACi) belinostat at clinically relevant concentration on nuclear factor erythroid 2-related factor 2 (NRF2) and mitochondrial metabolism for the treatment of KRAS-mutant human lung cancer. LC-MS metabolomic study of belinostat on mitochondrial metabolism was performed in G12C KRAS-mutant H358 non-small cell lung cancer cells. Furthermore, L-methionine (methyl-13C) isotope tracer was used to explore the effect of the belinostat in one carbon metabolism. Bioinformatic analyses of metabolomic data were performed to identify the pattern of the significantly regulated metabolites. To study the effect of belinostat on redox signaling Are-NRF2 pathway, luciferase reporter activity assay was done in stably transfected HepG2-C8 cells (containing pARE-TI-luciferase construct), followed by qPCR analysis of NRF2 and its target gene in H358 and A549 cells. Metabolomic study reveals significantly altered metabolic levels related to redox homeostasis including tricarboxylic acid (TCA) cycle metabolites (citrate, aconitate, fumarate, malate and α-ketoglutarate); urea cycle metabolites (Arginine, ornithine, arginino-succinate, aspartate and fumarate); and glutathione metabolism pathway (GSH/GSSG and NAD/NADH ratio) after belinostat treatment. 13C stable isotope labeling data indicates potential role of belinostat in creatine biosynthesis via methylation of guanidinoacetate. Moreover, belinostat downregulated the expression of NRF2 and its target gene NAD(P)H:quinone oxidoreductase 1 (NQO1), indicating potential anticancer effect of belinostat is mediated, potentially via Nrf2-regulated glutathione pathway. Another HDACi, panobinostat showed potential anticancer effect in both H358 and A549 cells.
Chapter 3 discusses the second application of metabolomics in pharmaceutical science. Herein, we utilized Pten knockout (KO)-induced prostate tumorigenesis mouse model to examine mechanism of action of phytochemical, indole-3-carbinol (I3C) via metabolic rewiring. I3C is a heterocyclic bioactive chemical found in cruciferous vegetables, possessing multiple biological activities such as antimicrobial, antioxidant, antiviral, anti-inflammatory, and anti-cancer activities. Untargeted metabolomics was performed using LC-MS based platform to investigate Pten-dependent and Pten-independent metabolic targets of I3C. The most impacted pathways by I3C included pyrimidine metabolism, arginine and proline metabolism, porphyrin metabolism, citrate cycle and lipoic acid metabolism.
Chapter 4 discusses the third application of metabolomics in pharmaceutical science. Herein, we established and verified an Nrf2 KO transgenic mouse model and studied resultant metabolic reprogramming in the gut and vital organs of mouse to understand the significant role of Nrf2 at metabolic level. To validate the Nrf2 KO mouse model, gene expression of Nrf2 and its target genes including Nqo1, Ho1 and Gclm were examined and were found to be significantly downregulated compared to wildtype. Additionally, regulation of antioxidant defense mechanism in the gut based on glutathione metabolism in small & large intestines of Nrf2 KO mice confirmed genetic knockout of Nrf2. Based on bioinformatic analysis of metabolomic information derived from small intestine, large intestine, liver, lung and brain of Nrf2 KO vs wild type mice, most regulated metabolites and metabolic pathways are identified and biologically interpreted. The most regulated metabolic pathways/metabolites in small intestine are citrate cycle and purine metabolism; in large intestine are pyrimidine metabolism; in liver are valine, leucine & isoleucine biosynthesis/degradation, phenylalanine biosynthesis/metabolism, fructose & mannose metabolism, taurine & hypotaurine metabolism, arginine & proline metabolism; in lung is lysine degradation; in brain are arginino succinate, cystine and xanthine. This has provided great starting point to further investigate correlation between metabolite concentrations and altered biological states mediated by Nrf2 in different mice tissues.
Chapter 5 provides conclusion for my research and discusses future perspectives including improved experimental design, bioinformatic and modelling approaches to further advance our understanding of metabolomic studies in biomedical research.
Appendix chapter 1 is a published review article that provides comprehensive review that identifies molecular biomarkers of pre-malignant and cancerous lung cells providing a strong basis to develop early cancer detection and interception methods. We discussed molecular changes, including genetic alterations, inflammation, signal transduction pathways, redox imbalance, epigenetic and proteomic signatures associated with initiation and progression of lung carcinoma. We also highlighted molecular targets of phytochemicals during lung cancer development. These targets mainly consist of cellular signaling pathways, epigenetic regulators and metabolic reprogramming. With growing interest in natural products research, translation of these compounds into new treatments or approaches to medical care has gained high importance. In this context, we discussed their overall disposition with pharmacokinetics. Lastly, we discussed clinical trials of phytochemicals in lung cancer patients.Ph.D.Includes bibliographical reference
Self-assembly of DNA nanostructures with organic drug ligands for lipid particle encapsulation
No full textChemotherapeutic drugs cause numerous side effects due to their mode of action, targeting both cancerous and healthy cells. This highlights the need for a more targeted delivery method. Under certain conditions, specific types of DNA sequences have the potential to self-assemble or adopt non-canonical three-dimensional (3D) secondary structures, such as G-quadruplexes and I-motifs. Similarly, four complementary DNA strands can self-assemble into Tetrahedral DNA Nanostructures (TDNs) engineered for drug delivery and diagnostics. This unorthodox pairing creates secondary structures capable of carrying organic drug ligands for delivery. Unlike TDNs, G-quadruplexes and I-motifs are therapeutic DNAs because they regulate gene expression and specific locations in the body, making them promising targets for anti-cancer research. Attaching a chemotherapeutic drug, such as Doxorubicin, to DNAs like G-quadruplexes and I-motifs—targets in anti-cancer research—should provide a synergistic effect, enhancing the drug's therapeutic ability. Encapsulating this DNA-drug complex in a lipid nanoparticle could further ensure more targeted delivery.
In this thesis, I investigated the formation of drug-loaded DNA nanoparticles under various conditions by using structures of G-quadruplexes, I-motifs, and tetrahedral DNA nanostructures. I have evaluated their capacity for drug delivery. A targeted delivery approach was explored to reduce side effects and combat drug resistance by encapsulating the DNA-drug complex within an ionizable lipid nanoparticle (LNP) for drug transport.M.S.Includes bibliographical reference