Environmental and Occupational Health Sciences Institute
Rutgers University Community RepositoryNot a member yet
58345 research outputs found
Sort by
A deep learning approach to the detection of illegal insider trading
No full textIllegal Insider trading involves the exploitation of non-public information for personal gain, posing risks to market fairness and investor trust. Tackling this issue is vital for ensuring transparency and equity in financial dealings. It is estimated that illegal insider trading is present in approximately 20% of merger and acquisition events and 5% of quarterly earnings announcements [1]. Our study introduces an innovative deep learning-based approach, leveraging a combination of text classification, time series forecasting and anomaly detection. By analyzing litigation releases from regulatory bodies and incorporating market and social media data, we develop a method to detect and predict signs of illegal insider trading. The addition of explainable AI techniques in our study offers clarity and transparency in the predictive models, making the rationale behind each decision more understandable.Our approach shows the potential of deep learning methods in identifying and providing insights into the unusual trading patterns concerning illegal insider trading and is a step towards more sophisticated monitoring tools. Our results demonstrate that the proposed approach effectively identifies potential instances of illegal insider trading, showing promise in detecting subtle patterns and anomalies that may indicate suspicious activity in financial markets. The findings suggest avenues for financial regulators and institutions to enhance their surveillance mechanisms, contributing to the ongoing efforts to maintain the fairness and reliability of financial markets.M.S.Includes bibliographical reference
Prediction of protein ampylation sites using machine learning
No full textProteins actively regulate cellular functions, and post-translational modifications like AMPylationare necessary for these regulations. AMPylation actively attaches adenosine monophosphate
(AMP) to a protein and influences cellular processes such as signal transduction, protein-protein
interactions, and enzymatic activity. Despite its importance, computational approaches for
AMPylation site prediction are still underexplored and do not receive the attention they deserve in
the broader scientific community. This thesis presents a machine learning-based framework for
predicting AMPylation sites in proteins, which compares traditional machine learning models with
deep learning approaches.
We extract features using the Position Specific Scoring Matrix (PSSM) and Structural Property
Descriptor 3 (SPD3) generated by SPIDER2. These yield 568 features, 400 bigrams from PSSM
and 168 from SPD3, with a window size of 10. We include 153 positive samples and 28,001
negative samples in the dataset and balance it using K-medoids clustering, which results in a final
feature matrix of 459 samples. We evaluate the models using 10-fold stratified cross-validation to
guarantee an accurate assessment.
As a result, Random Forest shows superior performance because it achieves an accuracy of
89.10%, a precision of 86.84%, and an F1 score of 80.54%. XGBoost achieves competitive results
and has an AUC of 93.86%. Although a Convolutional Neural Network (CNN) implemented using
Keras achieves comparable specificity and recall, the limited dataset size constrains its
performance and highlights the need for larger datasets to fully harness the potential of deep
learning.
This study creates a reliable framework for predicting AMPylation sites and emphasizes the
robustness of ensemble machine learning models like Random Forest in handling imbalanced
biological datasets. The findings open the door for more advanced predictive tools that help us
understand AMPylation’s biological importance and its broader role in cellular processes.M.S.Includes bibliographical reference
Supersymmetric Shimura operators and interpolation polynomials
No full textIn this dissertation, we give a partial super analog of a result obtained by Siddhartha Sahi and Genkai Zhang relating the Shimura operators and certain interpolation symmetric polynomials. We introduce the basic notions and give a survey of previous results in related areas. In particular, we study the pair (g = gl(2p|2q), k = gl(p|q)⊕gl(p|q)), define the supersymmetric Shimura operators in (U(g))^k, and prove that their images under the Harish-Chandra homomorphism are equal to specializations of Sergeev and Veselov's Type BC interpolation supersymmetric polynomials. This is our main theorem. We obtain results related to Harish-Chandra isomorphisms and generalized Verma modules along the way. Specifically, we show that it suffices to study certain elements in the center of U(g) and their actions on certain generalized Verma modules. We also offer a different approach to show the main theorem under the assumption that a family of irreducible g-modules is k-spherical. We introduce the notion of quasi-sphericity. With this, we prove the sphericity of these modules with the help of Kac modules in the case p=q=1. Many of the author’s results in this dissertation also appear in two earlier papers.Ph.D.Includes bibliographical reference
Improving the electrocatlytic activity of cobalt oxide for acidic oxygen evolution reaction with bismuth dopants
No full textThe continued overuse of fossil fuels for energy purposes does not only exacerbate the environment pollution but also causes the depletion of these energy resources. This makes the development of sustainable energy sources, such as green hydrogen, urgent and essential. Currently hydrogen is produced through traditional fossil fuel reforming or dehydrogenation processes. These processes, however, produce greenhouse gases and deplete fossil fuels. In contrast to these processes, water splitting utilizes an abundant raw material (i.e., water) and generates less pollution, if any. As such, this method exhibits a significant potential for massive production of hydrogen and aligns well with the principles of green chemistry. The method is especially quite appealing to produce green hydrogen when coupled with sustainable energy sources such as solar and wind energy. However, the oxygen evolution reaction (OER), a key reaction required for water splitting, hinders the overall efficiency of water splitting due to its slow reaction kinetics and high overpotential resulting from its intricate four-electron transfer processes. Furthermore, in acidic electrolyte, where the abundance of protons results in an efficient hydrogen evolution reaction (HER), most catalysts are unfortunately inactive and highly unstable towards OER. Thus, constructing stable, highly active acidic OER catalysts is of vital importance.
With this aim in mind, and to solve the issues mentioned above, a kind of high-performance acidic OER catalyst that is composed of bismuth-cobalt oxide annealed on fluorine-doped tin oxide (FTO) glass substrate is designed and synthesized. In the design of the catalyst, fundamental principles are considered. Co is included as it conforms with the so-called Sabatier principle well, which entails not only high catalytic activity, but also good stability in catalysts. Meanwhile, Bi is chosen as it is expected to work as a catalytically active center when it is with Co for OER. It is also expected to improve the density of active sites and oxygen vacancy in the catalysts, which are favorable for OER.
Additionally, catalysts containing Co and Bi in different ratios are synthesized to investigate the effects of their compositions on their electrocatalytic properties. The results show that Co9BiOx, which has a Co:Bi ratio of 9:1, exhibits the best electrocatalytic activity with the lowest overpotential (540 mV) at a current density of 10 mA·cm-2 and the lowest Tafel slope (113 mV·dec-1) to drive the OER. Its electrocatalytic activity is also found to be higher than those of the pristine Co3O4 and the catalysts prepared with other Co:Bi ratios. The high activity of the Co9BiOx, catalyst is accounted by its high electrochemical active surface area and double-layer capacitance. The participation of Bi in the electrocatalytic activities of the catalysts is further confirmed by density-functional theory (DFT) calculations.
This project’s findings not only highlight the potential of bismuth doping cobalt oxide to improve the acidic OER activity of this transition metal oxide, but also show the necessity of deeper investigations of p-block elements like bismuth for electrocatalytic applications. Furthermore, the research work leads to an eco-friendly catalyst for sustainable energy solutions.M.S.Includes bibliographical reference
Barriers, promoters, and predictors of cardiorespiratory and physical fitness among volunteer firefighters enrolled in the Cancer assessment and prevention study
No full textBackground: Volunteer firefighters who constitute a substantial portion of the US fire service, face significant challenges in maintaining optimal physical and cardiorespiratory fitness (CRF). This can impact their firefighting performance and increase the risk of health issues, including injuries, cardiovascular disease, and certain cancers. Despite their crucial role in protecting communities, volunteer firefighters face unique organizational and individual level barriers to obtaining or maintaining optimal physical fitness.
Objective and Specific Aims: This dissertation sought to investigate the prevalence, predictors, and influencers of physical fitness and CRF among volunteer firefighters enrolled in the Cancer Assessment and Prevention Study (CAPS), a sub-study of the national Fire Fighter Cancer Cohort Study (FFCCS). Three specific aims supported this goal: (1) investigate the prevalence and predictors of CRF among volunteer firefighters at different stages of their firefighting service and compare them to career firefighters, as well examine volunteer firefighters’ CRF in comparison to the general population of their respective state; (2) assess whether CRF levels among volunteer firefighters vary by fire department- and individual-level factors ; and (3) describe volunteer firefighters’ perspectives on how individual- and fire-department-level factors influence their physical fitness.
Methods: This dissertation included 569 incumbent volunteer firefighters. Their fire service, demographic, health, and occupational data were collected through an enrollment survey, and CRF was calculated based on self-reported age, sex, BMI, and physical activity rating. Study 1 compared volunteer firefighters to 1,874 incumbent career firefighters in the FFCCS, assessing fitness, health behaviors, and demographics. Mean CRF associations were analyzed through multivariable linear regression with an interaction term for years of service. CAPS volunteer firefighters' CRF and health behaviors were compared to state-specific populations using the Behavioral Risk Factor Surveillance System data. In Study 2, CAPS volunteer firefighter enrollment data were linked to information from the CAPS fire department survey, providing fitness and organizational details for 515 CAPS volunteer firefighters in 34 departments. The analysis explored associations between CRF and individual-level current fire service characteristics, as well as fire department characteristics, using mixed multivariable linear and logistic regression models, with fire department region included as a random intercept. In Study 3, 28 CAPS participants were interviewed individually, or in small groups based on their years of firefighting service, using an interview guide. The questions focused on participants' perceptions of their physical fitness in relation to their lifestyle and the department's impact. Thematic analysis and systematic coding were employed to analyze the interview transcripts.
Results: Study 1 found that on average volunteer firefighters had lower CRF than career firefighters and CRF among both groups was lower with more years of firefighting, with a greater magnitude among career firefighters. Study 2 showed that volunteer fire departments had a low prevalence of fitness related resources. Also, more frequent fire training was positively associated with volunteer firefighter CRF. In Study 3, five major themes emerged: (1) health and firefighting performance, (2) firefighter time and availability (3) departments’ responsibility to support members’ fitness, (4) fire training drills as a form of functional physical activity (5) fitness initiatives at the department.
Conclusion: The findings from this dissertation contribute to the health and fitness literature among volunteer firefighters, an understudied majority of the US fire service. The results also provide evidence that this population has limitations with maintaining or improving their fitness. The findings also inform health or fitness strategies to design and create tailored interventions for volunteer firefighters and departments.Ph.D.Includes bibliographical reference
Understanding the role of VR first-person perspective in climate change knowledge outcomes and pro-climate social actions
No full textConsidering the potential of virtual reality (VR) in addressing social change issues, a growing body of studies highlights the need for VR content production to address abstract and complex issues of climate change. In particular, people who feel undecided or uncertain about climate change because they cannot see the environment changing first-hand may become more engaged by a VR first-person perspective. Guided by the experiential media (EM) theory, particularly the first-person perspective quality of the EM, this dissertation explored the role of VR first-person perspective in climate change communication using a between-subjects experiment with a post-test only survey followed by an open-ended qualitative interview. It examined the effects of VR first-person perspective in climate change communication (Condition 1) on climate change knowledge, hope, fear, self-efficacy, believability of the VR experience, embodiment of the virtual avatar, and intended pro-climate actions compared to VR non-first-person perspective climate change communication (Condition 2) and VR first-person non-climate control condition (Condition 3).
This dissertation particularly explored RQ1) What is the effect of VR first-person perspective in climate change communication (compared to non-first-person climate change VR and VR first-person unrelated stimulus) on the dependent variables of climate change knowledge, embodiment, believability, hope, fear, self-efficacy, intended pro-climate social media actions and intended pro-climate offline actions? RQ2) From a qualitative standpoint, how do participants from first-person perspective in climate change communication (compared to non-first-person climate change VR and VR first-person unrelated stimulus) explain their VR experience in terms of the a.) what aspects of the content that stood out the most to them and what aspects they paid attention to, b.) embodiment of the virtual avatar, c.) believability of the experience, d.) hope and fear pertaining to climate change, and e.) intentions for engaging in pro-climate actions? Finally, this dissertation also involved a longitudinal survey component conducted after one month of VR exposure in order to see if the participants retained knowledge about climate change and whether participants in each VR condition actually did something pertaining to pro-climate actions one month after experiencing VR content production. The findings per the quantitative, qualitative along with the longitudinal data suggest mixed support for RQ1 and RQ2 in terms of the effects of VR first-person perspective experience of climate change communication on climate change knowledge and pro-climate social media and offline action compared to the VR climate non-first-person condition and the non-climate control condition. Particularly per the RQ1, the findings suggest that both climate change VR conditions (Condition 1 and Condition 2), regardless of first-person perspective, increased climate change knowledge and believability relative to the non-climate-related control condition (Condition3), thereby showing that the presence of first-person perspective did not improve knowledge acquisition or improve the believability of the VR experience. However, in terms of the embodiment of the virtual avatar, the climate change VR first-person condition (Condition 1) along with the VR first-person non-climate-related control condition (Condition 3) increased participants’ perceived embodiment of the virtual avatar relative to the VR non-first-person condition (Condition 2). Hence, the presence of first-person perspective did improve the embodiment of the virtual avatar. Further, the findings highlight that there was no statistically significant effect of the experimental treatment on hope, fear, self-efficacy, participants’ intentions for pro-climate social media actions, and intentions for pro-climate offline actions – as the mean levels were relatively similar across the three conditions. Further, the results of the longitudinal survey found that there were no significant differences between conditions on any of the variables tested i.e., climate change knowledge, hope, fear, and self-efficacy. Further, the average climate change knowledge levels did not vary significantly between conditions as they did on the immediate post-test; and the levels of hope, fear, and self-efficacy were comparable both on the immediate post-test and one month later, again with no differences between conditions. In terms of the RQ2, the findings per the qualitative open-ended questions suggest that the VR climate conditions, regardless of the first-person perspective, potentially helped in learning about climate change as participants in both conditions (Condition 1 and Condition 2) highlighted paying attention to various aspects of the climate change VR content and found various climate change scenarios stood out the most to them. The findings further suggest that regardless of first-person perspective, participants expressed feeling of hope pertaining to climate change relative to the non-climate-related control condition, with the difference being participants in Condition 1 felt seeing first-hand the comparison between a heavily polluted and barren landscape to a world with a clean and healthy environment with a lush green ecosystem achievable through climate-friendly actions hopeful alongside seeing 3-dimensional visualization of small climate-friendly actions and lifestyle scenarios as realistic and achievable to address the climate crisis, on the contrary, participants in the Condition 2 felt learning through the narrated and visual components about how daily lifestyle choices positively impact the environment and address climate change as feeling hopeful. It is also worth highlighting that participants in Condition 1 also expressed seeking hope in times of doomism. Next, the qualitative findings suggest that Condition 1 had an effect on participants' feeling of fear pertaining to climate change relatively higher than those in Condition 2 and Condition 3, and the climate first-person condition had a relatively higher effect on participants' believability in the VR experience than those in Condition 2 and Condition 3. Next, similar to the quantitative standpoints, the qualitative interview highlights that the climate first-person condition increased participants' perceived embodiment of the virtual avatar relative to the VR non-first-person condition, thereby confirming that the presence of the first-person perspective did improve the embodiment of the virtual avatar. On the contrary, regardless of the first-person perspective, participants expressed increased self-efficacy to fight climate change, suggesting that the presence of the first-person perspective did not improve self-efficacy. Likewise, regardless of the first-person perspective, participants expressed increased intentions for engaging in pro-climate offline and social media actions, suggesting that the first-person perspective did not increase intentions for pro-climate actions. In terms of the longitudinal survey results pertaining to what pro-climate offline and social media actions participants actually took after one month of the VR experience, the findings showed that participants in Condition 1 took most of the pro-climate social media and offline actions they had indicated in the post-test survey, except signing petitions to spread awareness about climate change, compared to the Participants in Condition 2 and Condition 3 who took relatively limited pro-climate social media and offline actions compared to what they had indicated in the post-test survey. Next, in terms of the plan to bring climate change to the attention of government and local political leaders, participants in Condition 1 offered a similar plan of action compared to their indication in the post-test survey with some additions such as participating in local community events that support climate change events, such as picking up trash. Further, one participant also expressed that they already have contacted the New Jersey State Senator to share the political leaders' views about climate change. Likewise, participants in Condition 2 also offered a similar plan of action compared to their indication in the post-test survey. However, the participants in Condition 3 offered plans that excluded specific actions, such as producing a video on climate change issues like trash in the river emission from factories, and sending the video to the government to bring climate change to their attention, and contacting the local political leader or government through traditional means of communication such as a phone call which the participants had indicated in the post-test survey, but excluded one month later in the longitudinal survey.
Finally, this dissertation offers suggestions for future research and practical implications, along with recommendations for developers of EM content and science communicators in the implementation of VR first-person perspective climate change projects at public schools, public libraries, or community centers to allow people of any age, race, gender, ability or social class, to transcend these categories and experience climate change realities in ways that might engender empathy and solidarity.Ph.D.Includes bibliographical reference
Formalized synthetic geometry
No full textWe offer a formalization of Book I of Euclid’s Elements in Lean4 (with tools fromMathlib4) using System E by Avigad, Dean, and Mumma. We contrast the proofs in this system to those of Euclid’s himself. We give detailed explanations for every theorem proved in Lean4 in order to complete Book I. Finally, we talk about the importance of tactics in Lean4 for the shortening and simplification of the proofs encountered in a geometric setting.Ph.D.Includes bibliographical reference
Camera communism: women photographers, avant-garde art, and documentary aesthetics in the Soviet Union, 1920s-1980s
No full textMany of the fundamental transformations in Soviet photography, and global modernist photography by extension, were made by women photographers. They challenged institutional privilege while pushing the boundaries of what was permissible in photography aesthetically, socially, and politically. Women did not merely enter the field during formative moments in photography’s history. Instead, they were active agents of that change, introducing to the visual field a different experience of living and working under socialism than their male peers. In reframing the influence of these photographers, many of whom have been ignored in scholarship on this topic, a new view of Soviet photographic history begins to emerge.
This dissertation is composed of five chapters that progress chronologically, with each one focusing on a photographer: Evgeniia Lemberg, Olga Ignatovich, Valentina Kulagina, Zenta Dzividzinska, and Natalia Tsekhomskaya. The first three chapters delineate the history between the 1917 revolution and World War II, looking at how female photographers navigated the emergence of the USSR’s system of state art. Chapter 1 introduces the frameworks and theoretical debates surrounding avant-garde, documentary, and Socialist Realist photography. It also investigates the brief career of Evgeniia Lemberg (1904-1934) and her working relationship with Aleksandr Rodchenko. Chapter 2 turns to Olga Ignatovich (1905-1984), focusing on her early career in the 1920s and 1930s to articulate the role of women in the emergence of photojournalism as a professional career path in the Soviet Union. Chapter 3 explores how photographers such as Valentina Kulagina (1902-1987) skillfully adapted avant-garde photomontage to the new aesthetic demands of Socialist Realism and Stalinism in the 1930s.
The last two chapters analyze how artists and photographers carved out space for dissent and artistic freedom between the 1960s and 1980s. Chapter 4 argues that early nonconformist photographers like Zenta Dzividzinska (1944-2011) were challenging established stylistic and institutional norms with approaches that echoed the radically experimental ethos of earlier counterparts. Chapter 5 focuses on the rise of nonconformist photography during the Soviet Union’s twilight decade of the 1980s through a case study of photojournalist and artist Natalia Tsekhomskaya (b. 1945). I argue that this last generation of Soviet photographers actively reexamined the groundbreaking work of earlier photographers by exploring new aesthetics that blended official dictates with nonconformist agendas. In this way, this dissertation traces how critical theories of photography intersect with radical political and artistic movements, to show how photography formed a discursive space for issues of gender equality in Soviet society.Ph.D.Includes bibliographical reference
Use of research and non-research inputs in policy deliberation: a longitudinal case study of communication in state policymaking regarding adolescent depression screening
No full textThis dissertation focuses on the relatively understudied role of communication in the deliberation phase of state-level policymaking, particularly in relation to the dynamic interactions between policy actors with a stake in policy implementation (e.g., experts, policy advocates, practitioners, and citizens) and policymakers navigating multiple sources of ambiguity regarding implementation. A conceptual framework is proposed for linking actors’ choice of inputs and communication strategy in the policy deliberation context to the likelihood of inputs to changes (minor or major) in the policy deliberated based on the integration of established communication, policymaking, and implementation science theories. A central focus of this framework is on explicating policymaker ambiguity in the specific context of decisions regarding policy implementation. Drawing on multiple conceptions of this construct in the communication and implementation science literature, the framework defines and operationalizes ambiguity as policymakers’ judgments or evaluations regarding the desirability and probability of implementation. Desirability is further conceptualized as the combination of perceived appropriateness and perceived acceptability of implementation. Probability of successful implementation is further conceptualized as the combination of perceived compatibility and perceived feasibility. The proposed framework is predicated on the assumption that inputs provided by policy actors in policy deliberations may inspire changes to policy implementation plans by either decreasing or increasing one or several critical dimensions of policymaker ambiguity. Accordingly, the research questions guiding the study are about the degree to which policy actors are strategic regarding their choice of inputs and communication strategies, the degree to which the content and focus of inputs they routinely provide in the policy deliberation context map onto the sub-dimensions of policymaker ambiguity explicated in the proposed conceptual model, and whether such strategic use of communication can inspire change to policy, and if so, under what circumstances.
These research questions, as well as the overall heuristic value of the proposed conceptual model, were explored via a case study analysis of the legislative process, from beginning (bill introduction) to end (bill signed into law), for a policy designed to improve universal access of adolescents to depression screening in schools in the state of New Jersey in line with current recommendations of the US Preventive Services Task Force and youth mental health expert panels. The case study utilized a multi-method approach (textual analysis of official documents and records related to this legislation and key informant interviews with policymakers and other policy actors directly involved in the policymaking process for this legislation) to track and analyze dynamic associations between inputs provided by different actors and changes made to the legislation as it progresses through the legislature. The research design implemented included three related components. Study 1 applied a valid and reliable document analysis research protocol and a measurement instrument to track and analyze use of research and non-research inputs by legislators and other policy actors in the context of school mental health policymaking in New Jersey to identify, characterize, and compare any stable patterns of evidence use (or evidence use routines) of different actors. This was done for the purpose of verifying key assumptions of the conceptual model (e.g., that actors are strategic about choice of evidence and the inputs they provide at different phases of the policymaking process) as well as for informing and contextualizing the more focused empirical investigation employed in Study 2. Study 2 focused exclusively on tracking and analyzing policy deliberations regarding the particular legislation that was the focus of the case study, probing associations among actors, the content and focus of the inputs they provide, and the communication strategies they employed in this context in line with the conceptual framework guiding the study. To corroborate, further contextualize, and triangulate findings emerging from these two studies, a series of key informant interviews were conducted with legislators and policy advocates with first-hand knowledge of the policymaking process for this legislation.
The findings across all three components of the study are largely consistent with the assumptions and rationale of the proposed conceptual framework, deeming it a promising target for future development and testing. Thus, there is sound evidence that policy actors regularly provide evidence-backed inputs throughout the policymaking process and particularly in the policy deliberation phase, when structured windows of opportunity open for actors both inside and outside the legislature to provide inputs. At the same time, use of evidence in this context is not monolithic, as different actors use different sources and types of evidence and different strategies for communicating evidence in this setting, and individual actors themselves may apply a variety of strategies, as a function of available resources and circumstances (such as degree of consensus among collaborating advocacy organizations regarding strategy).
Next, the findings emerging from both the document analysis and key informant interviews are consistent with the assumption underlying the proposed conceptual model regarding the centrality of policy ambiguity judgments as a mediator of the effect of inputs on policy. The findings of multiple data analyses demonstrate that inputs targeting sub-dimensions of ambiguity regarding policy implementation are associated, to various degrees, with changes to policy (whether minor/incremental or major/substantive), with clear implications regarding policy implementation. In this sense, the findings of this dissertation validate the proposed conceptual model and invites further theoretical development and empirical testing of hypotheses.
It is less clear how an actor’s choice of a communication strategy to deliver inputs may be associated with the likelihood of change to policy, in part because policy actors may not be considering this aspect in the context of their advocacy efforts. Of the four potential communication strategies included in the conceptual framework, which were theoretically derived from the strategic communication literature (education, persuasion, pressure, and negotiation), the findings provide significant support to education—and to a degree, to negotiation—as common strategies, with some findings (primarily from the key informant interviews) hinting at pressure as a strategy implemented via coordination of inputs among advocates. One potential practical implication of this is that policy advocates and other intermediaries operating at the science-policy nexus can benefit from adopting (or being trained on the application of) strategic communication planning tools that can increase their ability to promote sound evidence-based policies.Ph.D.Includes bibliographical reference
Confronting the divine in miniature: engraved gems and the creation of religious experience in Pompeii
No full textThis thesis explores how engraved gems from Roman Pompeii created religious experiences for their beholders. To this end, I analyze a corpus of examples with non-narrative images of deities comprised of 107 objects from the ancient city. Miniature in size and diverse in material, these objects have historically been considered separately from their context as worn objects which were physically and emotionally tied to their wearers. Drawing from theoretical frameworks that emphasize beholder experience and material agency, this thesis investigates how imagery, size, context, and media interact to create connections between wearer and the divine. I also consider comparative depictions of the gods in other contexts, such as shop facades and domestic altars, to explore how similar imagery interacted with viewers. Unlike previous investigations of engraved gems, this thesis emphasizes the human body as context for these objects, considering how their status as miniature worn objects transform the role of the image, in tandem with their materiality. In doing so, I suggest that the objects represented in the corpus created a religious experience, and further, that they cannot be understood without considering their connection, physical and personal, to wearers.M.A.Includes bibliographical reference