Environmental and Occupational Health Sciences Institute
Rutgers University Community RepositoryNot a member yet
58345 research outputs found
Sort by
Generating alpha through deep learning-based portfolio construction: bridging factor linkages with macro events in financial markets
No full textIn this study, we explore the efficacy of non-parametric methods in estimating conditionalasset pricing models, leveraging techniques such as LASSO and deep neural networks on
extensive conditioning information. We incorporate time-varying conditional information
related to alphas and betas derived from firm-specific attributes. We hypothesize that the
estimated alphas will remain stable over a one-month holding period, and we use these
alphas to construct portfolios. Additionally, the estimated alphas are selected through a
False Discovery Rate (FDR) process to ensure robustness. Beyond constructing portfolios
based on these estimated alphas, we employ the bootstrap method to identify more robust
short-term alphas.
Our asset pricing model demonstrates superior performance compared to benchmark
approaches in terms of Sharpe ratio, explained variation, and pricing errors. It also excels
in identifying key factors driving asset prices in out-of-sample evaluations. To further our
understanding of machine learning predictions, we analyze neural predictor performance
using the CRSP dataset contextualizing these findings with significant macroeconomic
events. The results reveal how neural networks distinguish and evaluate firm characteristics from the perspectives of risk and mispricing.Ph.D.Includes bibliographical reference
Additive manufacturing with a climbing mobile platform – design & implementation
No full textFused Filament Fabrication (FFF) is an additive manufacturing technology that deposits layer after layer of a material to build up objects over time. Since this technology offers significant advantages, including the ability to produce lattice structures and decrease wasted material, there is strong interest in producing meter-scale parts using FFF printers. However, simply expanding the size of a traditional gantry introduces high costs and requires extremely long print times due to the use of a single printhead. To solve these problems, mobile robot swarms capable of printing multiple distinct sections of an object simultaneously have been developed. But current solutions are limited in their dimensional capabilities since the mobile robot is always rooted to the ground. This work develops a printer attached to a mobile platform robot with the ability to climb onto previously deposited material and continue the printing process without the above dimensional limitations. A full process involves decomposing large objects into “blocks” that are printed when the mobile robot is stationary, followed by a maneuver to the next position and orientation so that the next block can be printed. It is demonstrated that with appropriate design of the robot paths this system is capable of printing a grid large enough for the robot to climb on top of previously printed layers and print the subsequent layers without letting the wheels fall off previous ones. Further, a smart-manufacturing motivated in-situ sensing capability is developed and demonstrated in which errors in inter-block alignment due to inaccuracies in the mobile platform’s movements are corrected by the printing platform.M.S.Includes bibliographical reference
Presumed cooperation as a pragmatic presumption
No full textThis dissertation investigates the role of perspectives in conversation, focusing on case studies ranging from locatives (‘left’), relative socio-cultural expressions (‘foreigner’), predicates of personal taste (PPTs like ‘tasty’), epistemic modals (‘might’), and epithets and thick terms (‘dyke’, ‘courageous’). I argue that a close look at patterns of constraints on felicitous use of such perspective-sensitive items (PSIs) undermines the assumption, common across philosophy of language, semantics, pragmatics, and cognitive science, that the speaker determines a conversation’s default perspective. To this end, I develop a notion of perspective that is able to adequately undergird the varied phenomena while also being sufficiently robust to play a role in utterance processing. I show that this notion of a perspective in hand with a clear understanding of what is meant by a processing default sets the stage for showing why a speaker default does not accord with much of the current experimental literature. Instead, I claim that the default is an extit{intersubjective} perspective, assumed to hold among all conversational participants. This hypothesis better explains data including generic readings of PPTs and modals, as well as slurs’ capacity to force tacit agreement. It also opens a broader program of research into the relationship among utterance comprehension, cognitive defaults, and conversational breakdown.Ph.D.Includes bibliographical reference
Adaptive convex loss selection for m-estimation: adaptive rate and efficiency
No full textThis thesis delves into the classical linear regression problem, focusing on the estimation of the
unknown parameter β0 ∈ Rd given n random vectors (X1, Y1), . . . , (Xn, Yn) that follow the linear
model Yi = X⊤i β0 + ϵi. The covariates X1, . . . , Xn are independent and identically distributed
(i.i.d.) Rd-valued and independent of the i.i.d. real-valued errors ϵ1, . . . , ϵn with an unknown
distribution. We focus on the low-dimensional setting where the problem dimension d is fixed,
analysing the performance of estimators as n becomes large.
Historically, the use of estimators that minimise the sum of the squared error losses has been
widely adopted in solving regression problems. In the linear regression setting, the Gauss–Markov
theorem justifies the ordinary least squares (OLS) estimator only under limited conditions. We be-
lieve that it is the convexity of the L2 loss function, which makes the minimisation computationally
tractable, makes OLS attractive to the practitioners. This thesis thus focuses on the adaptive (data-
driven) selection of convex loss functions for better estimation in linear regression, in the hope of
challenging the default use of L2 and L1 loss functions in modern high-dimensional regression.
Chapter 1 explores the mean estimation of univariate distributions symmetric around their mean
under irregular error distributions, where rates faster than n−1/2 can be achieved. We construct an
estimator achieving optimal rates up to n−1/α for compactly supported error distributions, where
α ∈ (0, 2]. The method involves a convex M -estimator based on a convex loss function Lγ (x) =
ii |x|γ , where γ is selected in data-driven fashion. The proposed method is computationally efficient
can be extended to linear regression settings. This chapter is based on the join work with Min Xu
and Cun-Hui Zhang, Kao et al. (2022).
Chapter 2 addresses linear regression under smooth error densities with finite Fisher informa-
tion, where a faster than n−1/2 rate is unattainable. We focus on selecting convex loss functions
to minimise asymptotic covariance among convex M -estimators. We identify the optimal convex
loss function with the log-concave projection of the noise distribution with respect to the Fisher di-
vergence and construct an adaptive estimator achieving the minimal asymptotic covariance among
the convex M -estimators. The proposed procedure is computationally efficient. This work is based
on the joint work with Oliver Y. Feng, Min Xu, and Richard J. Samworth, Feng et al. (2024).
iiiPh.D.Includes bibliographical reference
Risk factors for social and emotional loneliness in community-dwelling adults 60 years and older: A systematic review
No full textObjective: The objective of this review was to identify the risk factors for social and emotional loneliness in community-dwelling adults 60 years and older. Methods The proposed systematic review was conducted in accordance with the Joanna Briggs Institute (JBI) methodology for systematic reviews of etiology and risk. Results of the systematic review Identifying the included studies
A total of 467 titles resulted from the combination of searches through several databases and additional sources. After duplicates were removed, 333 studies were screened by the title and abstract for eligibility. Of the remaining articles, 20 full-text articles were retrieved, and each was assessed for eligibility. Ten articles were excluded as they did not meet the inclusion criteria. In general, studies with ineligible designs or those with ineligible outcomes were excluded. The remaining ten full-text studies were critically appraised and included in this review. Major implications for practice and research Practitioners should screen older adults for important identified risk factors such as being female, single, limited financial resources, poor mental health, and poor physical health. To address important risk factors for social and emotional loneliness strategies such as social engagement programs, community outreach, mental health support, intergenerational programs, peer support groups, addressing mobility issues, promote physical activity, family and caregiver support, policy and advocacy, education and awareness, and access to health services. More studies should assess specific interventions that address risk factors for loneliness.D.N.P.Includes bibliographical reference
Implementation of a Nurse-Driven Early Mobility Protocol
No full textPurpose of Project: Postoperative care management is important in optimizing patient outcomes and reducing the occurrence of postoperative complications. Early patient mobilization during the postoperative period has been linked to reducing patient length of stay (LOS) and the development of postoperative complications. This project aimed to investigate if implementing a nurse-driven mobility tool and a change in nursing perception would cause timelier postoperative patient mobility. Methodology: This quality improvement project was conducted at a large 738 academic medical center. There were two study populations for this project: patients who had undergone hepatobiliary surgery admitted directly from the post-anesthesia care unit to the surgical stepdown unit and nurses who are currently employed in the surgical stepdown unit. A total of 27 patient data was collected through a chart review, retrospectively and prospectively, looking into the primary outcomes of type of hepatobiliary surgery, LOS, what postop day the patient was first mobilized, development of postop complication, and what type, if applicable. A total of 20 nurses from both pre- and post-implementation participated in the anonymous online survey used to gauge nursing perceptions and beliefs on patient mobility. Descriptive and independent T-test statistics were used to analyze the project results via SPSS Statistics. Results: The mean LOS in days for patients in the pre-implementation was 7.07; the mean LOS in days for patients in the post-implementation was 4.89. During the pre-implementation period, 40% of the surgical patients had developed a postoperative complication as compared to 25% in the post-implementation period. There was a statistically significant in the LOS with the development of postoperative complications (p>0.001). There were no significant changes between the pre-and post-implementation nursing responses from the survey. Implications for Practice: This project showed there was not a lack of knowledge about the importance of patient mobility from the nurses, but a lack of confidence in mobilizing patients and structure from management. It would be beneficial for the institution to invest in more nursing education on safe patient handling and mobilization in collaboration with physical therapy and include patient mobilization as part of quality measures.D.N.P.Includes bibliographical referencesIncludes vit
Implementing a culturally tailored intervention to improve hypertension management among West African immigrants
No full textPurpose of the Project: Hypertension is a significant public health concern among West African immigrants in United States, with prevalence rates higher than the general U.S. population. However, cultural barriers and limited access to healthcare services can hinder effective management of the condition. This project aimed to evaluate the effectiveness of teaching hypertension through culturally appropriate, evidence-based teaching sessions to increase their awareness of the condition and motivate them to make positive lifestyle changes. Methodology: This quality improvement project utilized a quasi-experimental design with preand post-tests. 21 African immigrants were recruited. Four-week hypertension self-management sessions were given. The setting was a community Mosque in Newark NJ. Hypertension
knowledge and lifestyle modification (HELM) scale and Hill-Bone Compliance to High Blood
Pressure Therapy (HBCHBPT) scale were utilized to gather information on participants' knowledge, and attitudes towards hypertension management. Wilcoxon signed rank tests and Descriptive statistics were used to analyze the data. Result: In HBCHBPT Scale, reducing sodium intake subscale increases from 6.81 ± 1.436 to 10.1 ± 1.044 (Z = -3.86, p <. 001). Appointment keeping subscale increased from 6.71 ± 1.231 to7.24 ± 0.995 (Z =-2.232, p <.05). Medication adherence subscale increased from 30.62 ± 2.765 to 34.05 ± 1.962 (Z = -4.056, p <.001). The HBCHBPT pre- and post-implementation were
44.14 ± 4.053. and 51.38 ± 2.924, respectively (Z =-4.048, p<.001). HELM Scale increased from
9.81 ± 4.053 at pretest to 3.33 ± 0.796 at posttest (p < .001). Implications: Partnering with faith communities can help to increase the reach and impact of health promotion activities, as these communities often have a high level of trust and influence over their members.D.N.P.Includes bibliographical reference
Modulation of adaptive motor behavior by midbrain inhibitory neurons
No full textMidbrain dopamine neurons are an essential component of the basal ganglia thatbroadcast learning and motor signals to the forebrain. The pedunculopontine nucleus
(PPN), a midbrain structure that is mainly known for its role in locomotion and arousal,
is among a handful of hindbrain regions that send heavy efferents to the dopaminergic
midbrain. Glutamatergic and cholinergic efferents from the PPN have been shown to
modulate the phasic activity of dopamine neurons in the substantia nigra compacta,
shaping their responses to prediction error signals, their changes in behavioral
contingencies and their activity during self-generated movement initiation. We recently
discovered the existence of a third PPN projection system that originates in inhibitory
neurons (GABA) and directly innervate dopamine neurons of the substantia nigra
compacta. Activation of PPN GABAergic axons inhibits dopamine neurons and stops
locomotion in freely moving mice. Given the role of dopamine neurons in adaptive
behavior and the role of the PPN in behavioral initiation, we designed a series of
experiments to test the hypothesis that PPN GABAergic neurons contribute to shape
dopamine neuron activity during goal-directed actions. In the first aim, we used an
optogenetic approach to activate PPN GABAergic axons in the SNc during an operant
task that required mice to complete a sequence of lever presses to get a reward. We
found that PPN activation caused the execution of ongoing motor sequences to pause,
while the same activation before a sequence is initiated, in some cases, aborted the
upcoming sequence. When PPN axons were activated in tasks that did not involve self-
generated motor bouts, we observed no detectable changes in the performance of
those actions. In the second aim, we investigated the role of PPN GABAergic signaling
during spontaneous movement (i.e. locomotion initiation, termination, execution speed)
by recording the activity of optogenetically-identified GABAergic PPN neurons. Acute
recordings were made from head-fixed mice that were free to run on a frictionless
wheel. Overall, PPN unit activity varied in their firing characteristics, as previous reports
have shown. We identified PPN neurons that are speed modulated, with varying
correlation strengths. These neurons also showed time-locked activity changes during
locomotor initiation and termination, but no phasic changes were observed. Identified
GABAergic neurons showed less overall modulation with speed, and were not recruited
during spontaneous locomotor initiation and terminations. In the third aim, we tested
whether PPN GABAergic neurons are recruited during the initiation of goal-directed
movements by training mice in a go/no go paradigm. Our results show that the PPN
GABA contributes to the difference between action representations in trials that mice
have responded to the conditioned stimuli, by specifically marking response
terminations, suggesting a role in behavioral switching. Implications of such a role are
discussed, as well as future directions.Ph.D.Includes bibliographical reference
A comparative analysis of ethnopolitical movement strategy: the Kurds and Pashtuns
No full textWhy do some minority-based social movements seeking autonomy or separatism respond to state repression with armed resistance while others respond with unarmed resistance? To answer the research question, I examine variations in strategies of resistance that have occurred across two structurally similar ethnopolitical struggles in Islamic countries: the struggle for Pashtun rights in Pakistan and the struggle for Kurdish rights in Turkey. Incorporating a longitudinal paired comparison methodology, the research explains how and why strategic shifts between armed and unarmed resistance have occurred over time among groups struggling for autonomy or separatism and how and why challengers engaged in unarmed resistance may or may not gain leverage in authoritarian contexts. The research findings should be generalizable beyond the two struggles and provide insights into the possibilities of civil resistance in autonomist and separatist struggles more generally. The research contributes to the comparative and historical study of ethnopolitical conflict and the literatures on social movements and ethnopolitics. Keywords: Kurds, Pashtuns, Turkey, Pakistan, ethnopolitical, nationalism, ethnonationalism, non(violence)Ph.D.Includes bibliographical reference
Three essays on voice-based social media: user behavior, technology adoption, and monetization strategies
No full textThis dissertation, composed of three research papers, explores the dynamics of voice-based social media platforms, such as Clubhouse and Twitter Spaces, focusing on the multifaceted dynamics of user engagement, technology acceptance, and monetization strategies. The first paper examines platform usage's social and psychological determinants, utilizing theories like Social Presence Theory and Social Identity Theory. Through a quantitative approach involving Partial Least Squares Structural Equation Modeling (PLS-SEM) and a survey of 340 respondents, it reveals that social identity, communication preferences, and perceived social support are key drivers of user engagement, while privacy concerns and communication frequency play lesser roles. The second paper investigates these platforms' technology acceptance and marketing potential by integrating the Technology Acceptance Model (TAM) with social media marketing constructs. Employing Structural Equation Modeling (SEM) and survey data from 303 active social media users, it identifies trendiness and word-of-mouth as significant predictors of behavioral intentions, with perceived value playing a mediating role while highlighting the need for comprehensive design and marketing strategies. The third paper delves into the factors influencing users' willingness to pay for premium features, using a survey of 351 users and quantitative analysis to develop a conceptual model based on the Theory of Planned Behavior (TPB) and social influence theories. It emphasizes the critical role of community engagement and perceived credibility in driving willingness to pay while noting the limited impact of social proof and scarcity. Together, these studies provide a robust, data-driven understanding of the factors that influence the adoption, usage, and monetization of voice-based social media platforms, offering valuable insights for developers and marketers in optimizing user engagement and capitalizing on the unique opportunities these platforms present in today’s digital landscape.Ph.D.Includes bibliographical reference