University of Maryland, Baltimore County
Not a member yet
17643 research outputs found
Sort by
Mathematical Models for the Etiology of Schizophrenia and White Matter Lesions
No full textThe thesis consists of two projects. The first project uses mathematical models from nuclear physics to explore epidemiological data related to schizophrenia. These models improve the state of the art understanding of the biological etiology of schizophrenia, suggesting that regular internal biological events are responsible for disease development. The schizophrenia project develops two families of mathematical models that describe the course of schizophrenia. First, the models are applied to schizophrenia prevalence data for different populations. Parameters from these models are analyzed for trends relating to the parameters. The parameters are used to simulate datasets showing the relationship of the models back to the observed parameters. These models from theoretical physics can explain monozygotic twin discordance in schizophrenia. The second project explores white matter lesions in a Mexican-American population across the adult lifespan. A novel mathematical model is created to relate white matter lesion development to aging, diabetes, and hypertension. The white matter lesion project examined real data from a Mexican-American population. The model revealed that diabetes, hypertension, and age are strongly associated with the development of white matter lesions. The data revealed a transition from lower volume, number, and average volume of lesions in the 36-45 to 46-55 decades of life. The novel mathematical model uses a logistic differential equation and elements of probability theory to recreate the data. Further analysis of the model showed that it not only fit the Mexican-American data, but also fit data related to the Austrian Stroke Prevention Study. It made predictions about the effects of diabetes and hypertension in a simulated Mexican-American population. The totality of the projects show that physics is a fertile ground for developing physically based mathematical models that can be applied to diverse problems relating to medicine. Potential extensions to this work will also be discussed
Learning Hierarchical Workflows Using Community Detection
No full textWorkflows identified from user event logs and click-stream data are useful as knowledge bases for behavioral analysis and recommendation systems. In this study we identify abstractions or summaries of event logs modeled as user activity flow networks. The abstractions are identified based on structural properties as well as user activity flow dynamics over the network using community detection methods. We apply a fast modularity optimization and multi-level resolution approach to detect hierarchical community structure in user activity flow networks. The detected communities are compared to those detected by the information-theoretic map equation minimization approach to weigh pros and cons of the fast modularity optimization approach in the workflows context. We further attempt to identify the most probable sources and sinks of user activity in individual communities and trim the network accordingly to reduce entropy of the workflow abstractions
SOMETIMES I WANT TO SHAKE THE BRANCHES OF MY FAMILY TREE AND SEE WHAT FALLS OUT
No full textSOMETIMES I WANT TO SHAKE THE BRANCHES OF MY FAMILY TREE AND SEE WHAT FALLS OUT is an installation piece exploring the personal archive as a medium for creating a historical narrative. The work uses three themes to explore the historical object and the archive; the real, the fake, and the scrutiny. All three themes are examined through the artworks of three female artists investigating these motifs and their relationship with history. The Branches are a line of personal ancestors whose recorded histories are easily available digitally. This piece is a work that studies the monolithic properties of the archive, but also its perpetual partiality, which constantly drives and dictates a specific historical narrative
Multivariate spatial anomalous window discovery
No full textThis dissertation investigates and formulates a novel multivariate spatial anomalous window discovery method. This novel method discovers linear shaped spatial anomalous windows corresponding to where unusual phenomenon take place. This method also uses a non- parametric multivariate scan statistic model that does not rely on any prior knowledge of the data distribution. Anomaly detection in spatial data has been a challenging research problem as most of the traditional anomaly detection approaches are not suitable to this work. In contrast, scan statistic approaches have proven to be a promising technique in quantifying spatial anomalous windows, where the anomalous window is a contiguous set of objects in a region forming an unusual cluster with respect to the rest of the data. Anomalous windows are found in several real world applications such as accident hubs along highways, disease outbreaks, crime hot spots to name a few. Existing scan statistic approaches have several limitations. First, they do not identify robust, multivariate, linear form, non parametric windows which are critical in studying phenomenon such as traffic accidents. Second, multivariate data may be sparse or noisy, thus, it is important to identify anomalies taking into account the presence of outliers in the data. Third, real world data does not necessarily follow known distributions, thus, it is important to identify anomalies in a non parametric setting such that it does not rely on specific fitted distributions in the data. Extensive experiments on multiple real-world data sets are conducted. The experimental results demonstrate the efficacy of our novel method in discovering critical anomalies in real-world applications. For example, the linear anomalous window discovery on real-world accident datasets for various highways with known issues were validated with existing domain expert reports to prove our results are not only statistically significant, but also identify the real anomalous traffic accident hubs along multiple intersecting highways
Predicting Stable and Cost-Effective Medicaid Nursing Home Transitions
No full textIn order to examine the characteristics of individuals who transition from nursing homes to the community, a group of 11,560 Medicaid participants aged 65 and older who had a nursing home admission between April 1, 2003, and April 1, 2006, were selected. Overall 231 of them transitioned to community; nearly three quarters of transitioned individuals wished to return to the community and over half had a support person who was positive toward discharge. Compared to all residents in the study, those who transitioned to the community were more likely to be younger, had a much lower average level of care, were in their Medicaid eligibility span for less than one year when they entered the nursing home, and were more likely to need lower levels of assistance with their activities of daily living. A Heckman selection model was used in order to determine if any factors exist that can identify nursing home residents who are most likely to transition to stable living in the community. The model found that there were several known factors that can predict transition from a nursing home to the community, including age, gender, marital status, level of care, preference to return to the community, and the presence of a support person positive toward discharge. The second stage of the model, however, failed to find any factors significantly associated with an individual being stable in the community, suggesting that post-transition episodes (such as a heart attack or fall), not pre-transition characteristics, are responsible for truncated community living. Finally, an endogenous switching regression was used to determine how much Medicaid cost savings is associated with transitioning to the community, and how that cost savings is affected by individual characteristics. By comparing transitioned individuals to a matched group of nursing home residents, this study reaffirmed prior findings that transitioning older adults from a nursing home to the community is associated with Medicaid cost savings, in this case of approximately $1,900 per month. In addition, the findings show that there is a known factor--pre-transition cost--that has a significant effect on post-transition Medicaid spending
Tensor-based Spatio-Temporal Outlier Detection in Large Datasets
No full textSpatio-Temporal data is inherently large since each spatial node has spatial attributes and may also be associated with large amounts of measurement data captured over time. In such large and multi-dimensional data identifying anomalies can be a challenge due to the massive data size and relationships among spatial objects. Discovering anomalies in spatio-temporal data is relevant in several domains, such as detecting rare disease outbreaks, detecting oil spills, discovering regions with highway traffic congestion. Most existing techniques for discovering anomalies in spatio-temporal data may find the spatial outliers first and then identify the spatio-temporal anomalies from the data of that specific spatial location. Alternatively, some approaches may discover anomalous time periods and then discover the unusual spatial location in them. This may lead to identifying incorrect spatio-temporal outliers or missing important spatio-temporal phenomena due to the elimination of information after each step. Thus, there is a need to address capturing both space and time simultaneously. A tensor is a multi-dimensional array. It is considered as a powerful tool to manipulate multi-dimensional and multi-variate data. It has a concise mathematical framework for formulating and solving complex data problems efficiently. Tensors can handle complex relationship in spatio-temporal data and their mathematical framework can help us detect spatio-temporal outliers in an effective and efficient manner. Tensors multiplication can integrate spatial and temporal aspects in the data at the same time. In this dissertation, we present our novel approach addressing the key limitation of existing spatio-temporal outlier detection methods by using an efficient tensor based model that supports complex relationships in spatio-temporal data to detect outliers by looking at space and time simultaneously as well as handling the scalability issue when it comes to manipulating large datasets. In this dissertation, we are going to present our novel spatio-temporal tensor model. Based on the spatio-temporal tensor model, we present our clustering-based neighborhood discovery algorithm, neighborhood-based spatial, and spatio-temporal outlier detection algorithms to discover different types of spatio-temporal outliers namely point based and window based outliers. We discuss detailed experimental results for each of the algorithms proposed and also present comparative results
Modeling and design of lossy waveguide structures for generation of broadband terahertz pulses through difference frequency mixing
No full textWe present an integral coupled mode theory (CMT), suited to account for high optical losses, to model ultra-broadband terahertz (THz) waveguide emitters (0.1- 20 THz) based on difference frequency generation (DFG) pumped by femtosecond infrared (IR) optical pulses. This integral model works even in the situation where the DFG occurs between several IR and THz modes. We also present a simplified CMT approximation that reproduces the results of the rigorous integral CMT for situations where the THz generation is mediated through single-IR-mode to single-THz-mode interactions. Using the simplified approach we derive a new expression that incorporates loss effects into the coherence length for optical rectification (OR). The expression that we derived for the coherence length can be adapted to describe other second order nonlinear processes such as second harmonic generation. We apply both models to study waveguide emitters whose nonlinear cores are composed of poled guest-host electro-optic (EO) polymer composites, which belong to the 1mm symmetry class and have high nonlinearities. We apply the models to a generic, symmetric, five-layer, metal/cladding/core waveguide structure and provide design strategies for efficient ultra-broadband THz emitters. Two different design strategies are analyzed, one in which the waveguides are designed to have a single-IR-mode and a single-THz-mode guided within the structure, and other where the waveguide is made with a single-THz-mode but admits several IR guided modes. In both strategies the waveguide geometric parameters are optimized to obtain the highest THz conversion efficiencies and broader output bandwidth. The simplified CMT approach is much faster to implement than the integral CMT. Thus, we use the simplified approach to perform a parametric study for different waveguide parameters and pumping wavelengths, in the telecom and short wavelength infrared region, to establish under what conditions the five-layered structure yields its highest conversion efficiencies. Coupling conditions are also optimized to guarantee that most of the incident pump power (~ 80%) is utilized in the interaction. We find conversion efficiencies as high as 35x10−4 W−1 and bandwidths up to 20 THz for a structure with a core made of EO polymer AJTB203, polystyrene cladding layers and Al metal-capping layers, when pumped at 1900 nm. We found that for an optimized structure there must be a perfect balance between both modal phase-matching and mode losses effects. Also, we observed that low-loss-cladding layers enhance the efficiency for phase-matched structures, increase the interaction length, and improve the stability of the efficiency with respect to variations in waveguide parameters. Finally, we identified under what conditions the simplified CMT approximation fails to describe broadband THz generation. For five-layered structures with thin cladding layers (~ 200 nm thick) both the fundamental and first excited IR modes are involved in the DFG interaction and both must be accounted for to completely describe THz DFG. This is necessary since the fundamental TM-IR mode of structure hybridizes into a plasmon mode as the cladding layers are thinned and THz generation occurs also through OR of the first excited even TM IR mode. The first excited even TM IR mode acts as a quasi-fundamental mode in the absence of cladding layers. In this case the full integral CMT formulation with must be used to correctly model the THz generation
ENGLISH LEARNERS WITH LIMITED OR INTERRUPTED FORMAL EDUCATION: RISK AND RESILIENCE IN EDUCATIONAL OUTCOMES
No full textThis dissertation examined the educational outcomes of high school English learner (EL) students with limited or interrupted formal education (SLIFE) to evaluate theories that explain their educational resilience. School system data and survey results from 165 high school ELs were analyzed to determine the degree to which ELs' homeland schooling had influenced their academic outcomes in the U.S. Educational outcomes included English proficiency attainment and gains as well as scores on standardized tests of algebra, biology, and English language arts. Limited formal schooling (LFS) was operationalized with three indicators for students on arrival in the U.S.: (1) gaps in years of schooling relative to grade, (2) low self-reported first language schooling, and (3) beginner-level English proficiency. Bivariate and multivariate regression analyses were used to estimate the relationships between the LFS indicators and the educational outcomes as well as the degree to which school-based protective factors and personal risk factors had influenced the relationships. Protective factors included perceived pedagogical caring, social integration with non-immigrant peers, ESOL classes, out-of-school help, and extra-curricular activities. Risk factors included high social distance, past traumatic experiences, a lack of authoritative parental support, separations from loved ones, and hours spent working in employment. This study also examined the role students' academic self-concept played in mediating and moderating the influence of protective and risk factors in the resiliency process. The findings showed that SLIFE had lower achievement on the standardized tests, but that it was largely due to having lower English proficiency at the time of the test. Lower English proficiency at the time of the test was mainly attributed to arriving with lower English proficiency and lower first language literacy. ESOL classes appeared to help students acquire English faster. After controlling for differences in English proficiency, students' perceptions of social distance appeared to predict their academic achievement on standardized tests better than their academic self-concept and the other protective or risk factors. This study contributes to our understanding of risk and resilience among SLIFE and may help inform interventions to support them better
HETEROTRIMERIC G-PROTEIN βγ SUBUNITS IN THE MOUSE PERIPHERAL OLFACTORY SYSTEM
No full textG-protein coupled signaling cascades are involved in a variety of activities in the sensory neurons of the peripheral olfactory system in mice, such as odor transduction and axonal projection. While the roles of heterotrimeric G-protein α-subunits, such as the olfactory G-protein α subunit (Gαolf) are well documented, very little is known about Gβγ subunits in the nasal sensory epithelium. Contrastingly, many studies in other systems have yielded a surge of data documenting the diverse functions of the known five Gβ and twelve Gγ subunits. Lin et al. in 2007 characterized a subset of olfactory sensory neurons in the main olfactory epithelium expressing the ion channel transient receptor potential channel (TRPM5). In other systems such as taste receptor cells, TRPM5 is a necessary downstream component of the phospholipase C (PLC) pathway which is thought to be activated by Gβγ. The heterotrimeric G-protein γ subunit Gγ13 is present in the cilia of olfactory sensory neurons (OSNs) of the main olfactory epithelium (MOE). This suggests that as an obligate dimer, Gβγ could be serving a role in the function of olfactory sensory neurons. Although Gγ13 expression was found in the MOE, there was scattered and incomplete data regarding expression of other β and γ subunits in both MOE and the vomeronasal organ (VNO). In this work, using molecular techniques I provide a complete expression profile of the 5 different Gβ and 12 Gγ subunits in the MOE and VNO. I show that multiple Gβ and Gγ subunits are expressed in the MOE and VNO in a cell type specific fashion. Additionally, using molecular techniques and image analysis, I show that Gβ1 is dominant in both MOE and VNO. Finally, using molecular and physiological approaches, I demonstrate that Gβ1 is essential to proper function of OSNs and also for glomerular formation in the main olfactory bulb (MOB). My results demonstrate the heterotrimeric G-protein βγ subunits serve critical roles in olfaction
Intimate partner violence recidivism over an 8-year follow-up
No full textThe study examined the criminal behavior of men who sought treatment for intimate partner violence (IPV). Criminal offenses were coded for the 5 years prior to treatment initiation and the 8 years after treatment initiation into three categories: Domestic Abuse, General Violence, and Other Protection Order Involvement. Of 115 men, 85 (73.91%) had no incidents coded as Domestic Abuse during the 8-year follow-up period. Overall, there were significant reductions in Domestic Abuse and General Violence between the two time periods. It was observed that 12 men (10.4%) accounted for 60% of the incidents of Domestic Abuse recidivism. The main study hypothesis was that borderline and antisocial personality characteristics would be predictive of criminal recidivism after treatment initiation. Counter to hypotheses, these personality characteristics were not predictive of criminal recidivism in the majority of analyses. Consistent with the hypothesis, self-report ratings of antisocial personality characteristics, assessed at treatment program intake, were significantly related to General Violence during the 8-year follow-up. When incidents of General Violence and Domestic Abuse recidivism were combined, this variable was also significantly related to previous self-ratings of antisocial personality characteristics. Self-reports of borderline personality characteristics were not significantly related to any of the criminal recidivism variables. The findings suggest that the majority of men seeking treatment for IPV have no criminal recidivism after treatment initiation, but that a small subsample of men maintain or increase the frequency of criminal behavior after treatment initiation. Future studies will need to examine other individual-level characteristics that may distinguish these men so that treatment can better target the risk factors of IPV recidivism