University of Illinois at Chicago
University of Illinois at Chicago: UIC INDIGO (INtellectual property in DIGital form available online in an Open environment)Not a member yet
21439 research outputs found
Sort by
Learning Agenda for Systems Change: A Toolkit for Public Health, Partners, and Communities
No full textThe LASC Toolkit 2.0 is an action-oriented, evidence-informed, field-tested process for organizations (as well as their partners and communities) to use transformative learning as a driver for equity-focused systems change.</p
<i>Linking Trials to Publications: Enhancing Recall </i><i>by Identifying Trial Registry Mentions in Full-Text</i>
No full textWe have developed a free, public web-based tool, Trials to Publications, https://arrowsmith.psych.uic.edu/cgi-bin/arrowsmith_uic/TrialPubLinking/trial_pub_link_start.cgi, which employs a machine learning model to predict which publications are likely to present clinical outcome results from a given registered trial in ClinicalTrials.gov. The tool has reasonably high precision, yet in a recent study we found that when registry mentions are not explicitly listed in metadata, textual clues (in title, abstract or other metadata) could identify only roughly 1/3 to1/2 of the publications with high confidence. This finding has led us to expand the scope of the tool, to search for explicit mentions of registry numbers that are located within the full-text of publications. We have now retrieved ClinicalTrials.gov registry number mentions (NCT numbers) from the full-text of 3 online biomedical article collections (open access PubMed Central, EuroPMC, and OpenAlex), as well as retrieving biomedical citations that are mentioned within the ClinicalTrials.gov registry itself. These methods greatly increase the recall of identifying linked publications, and should assist those carrying out evidence syntheses as well as those studying the meta-science of clinical trials.</p
Towards Subpolynomial-time Fully Dynamic Minimum Cut
No full textIn the age of big data, traditional static graph algorithms struggle to keep up with the
dynamic nature of real-world networks. Dynamic graph algorithms become increasingly important
because they allow for the continuous updating and querying of graphs as new data comes
in, enabling real-time insights and decision-making. This capability is crucial for real-world
applications where timely information can significantly impact outcomes.
In this dissertation, we present fully dynamic algorithms for the graph c-edge connectivity
problem. We begin by discussing the history of fully dynamic edge connectivity research and
the improvements made in the last decade that our research builds upon.
We then present a deterministic fully dynamic data structure that can answer c-edge
connectivity queries. Given a dynamic graph G with n vertices that undergoes edge insertion
and deletion updates, our data structure is able to answer c-edge connectivity queries on a pair
of vertices in no(1) time worst-case update and query time, for any c = logo(1) n. This result
was published in FOCS 2021.
Next, building upon our previously described c-edge connectivity data structure, we develop
an extended dynamic data structure that deterministically maintains a minimum cut of size at
most c, where c = logo(1) n. This new data structure achieves no(1) worst-case time complexity
for both updates and queries. Our work on this topic was presented at SODA 2024.
Finally, we discuss the barriers and attempts to remove the c = logo(1) n qualifier
The Day-to-Day Bidirectional Association of Sleep Difficulties with Affect in Postmenopausal Women
No full textPostmenopausal women are at elevated risk for sleep difficulties and changes in positive affect (PA; energy, interest, etc.) and negative affect (NA; anger, stress, etc.) due to numerous factors, including vasomotor symptoms (VMS). In the general population, there is evidence for a day-to-day bidirectional association between sleep difficulties and affect, such that sleep difficulties predict next day affect and affect predicts next night sleep difficulties. However, this association has not been examined in postmenopausal women, a population susceptible to both sleep difficulties and changes in affect. Therefore, we examined the day-to-day bidirectional association between sleep difficulties and affect in postmenopausal women, and further examined if the relationship differed by the presence of VMS.
Participants were enrolled in MsBrain I. Both sleep difficulties and affect were assessed over a 72-hour period at home. Sleep difficulties were assessed via a wrist accelerometer and sleep parameters of interest included total sleep time (TST), sleep efficiency (SE), and wake after sleep onset (WASO). Affect was assessed via ecological momentary assessment. Participants were prompted four times per day on a handheld device to rate how strongly they currently felt each emotion (PA: energetic, interested/involved; NA: angry/irritated, nervous/anxious, and stressed). Total PA and total NA scores were calculated from each survey response and a daily average total PA score and a daily average total NA score was calculated using responses to each survey. VMS were assessed over 24-hours using a sternal skin conductance monitor. Women with objectively measured VMS were labelled as Flashers and women without VMS were labelled as Non-Flashers. Associations were evaluated using mixed-effects models with a random intercept for participant and autoregression process of order one. All models controlled for age, race, years of education, income, marital status, night shift work, sleep medication use, and anxiolytic medication use (Model 1). Depressive symptoms (Model 2) and 24-hour VMS frequency (Model 3) were added as covariates in a hierarchical manner.
Participants included 247 postmenopausal women (average age = 59.27, 81.38% White, 77.73% Flashers) with two or more days of concurrent sleep and affect monitoring and 24-hours of valid VMS monitoring. Sleep difficulties predicted next day affect such that moderate TST, higher SE, and lower WASO predicted higher next day PA. Sleep difficulties were unrelated to next day NA. Affect predicted next night sleep difficulties such that higher PA predicted shorter TST and lower WASO. PA did not predict next night SE and NA was unrelated to next night sleep difficulties. VMS status did not moderate the day-to-day bidirectional association between sleep difficulties and affect in all but one case. In that case, moderate TST predicted higher next day PA in Non-Flashers and TST was unrelated to next day PA in Flashers. All associations in Model 1 persisted in Model 2 and Model 3.
Findings support a day-to-day bidirectional association between sleep difficulties and PA in postmenopausal women, with better sleep associated with higher next day PA and higher PA associated with changes in next night sleep. In all but one case, this association did not differ by the presence of VMS. These results suggest that targeted therapies for sleep improvement (e.g., cognitive behavioural therapy for insomnia, etc.) may enhance next day positive emotions and targeted practices for increasing positive emotions (e.g., yoga, etc.) may alter next night sleep in postmenopausal women with and without VMS
AI-Driven Indoor Environmental Monitoring for Smart Buildings Using Graph Neural Networks
No full textAir pollution has been identified as one of the world’s leading environmental health risks, impacting millions globally. Ensuring protection against its harmful effects has become a pri- mary concern, especially in indoor environments where people spend the majority of their time. This thesis presents a novel framework focused on the estimation and spatial interpolation of PM2.5 concentrations within indoor spaces, enhancing the capability to monitor and mitigate air pollution in real time. Our approach combines recurrent and feed-forward deep learning models, specifically leveraging a Graph Neural Network (GNN) model that captures spatiotem- poral dependencies and integrates contextual features unique to indoor environments. The model is engineered to process sensor data from different indoor locations while considering complex airflow obstacles, including walls, HVAC vents, air purifiers, and their operational statuses.
Furthermore, this framework can be extended to monitor additional environmental param- eters within indoor spaces, such as PM10, CO2, temperature, and noise, broadening its utility in creating healthier indoor environments and saving costs for facility managers. Finally, we propose a scalable cloud-based deployment strategy to support efficient, real-time monitoring and enhance integration across multiple client sites
Predictive Modeling of Application Runtime in Dragonfly Systems
No full textThe Dragonfly interconnect is widely adopted by extreme-scale systems, yet its sharing nature often results in trac from various applications competing for network resources, causing workload interference and leading to variable application runtime.
This work aims to leverage deep neural network methods to forecast application iteration times, using network features collected at the router port level.
The problem is addressed by employing graph neural network-based dynamic models that are trained on an ad-hoc graph structure that reflects the physical characteristics of the system, and can capture its temporal and structural dynamics.
Results show that this methodology is able to outperform the baselines for one and two future steps ahead. However, it faces scalability challenges when applied to larger systems. To address these limitations, the methodology was enhanced by constructing an ensemble model that integrates a custom GNN-based component with the recently proposed TimeLLM framework, which leverages large language models for time series forecasting
Experimental and Computational Studies of Hierarchically Organized Nanomaterials
No full textThis thesis investigates the controlled self-assembly of complex hierarchical materials across multiple length scales, focusing on their design, properties, and applications. A significant emphasis is placed on the development of novel, reversible chiral superstructures based on metal-amino acid building blocks. These materials exhibit tunable physicochemical and chiroptical properties, making them promising candidates for applications in chiroptical devices, biomimetic materials, and chemical separation techniques. Their reversible nature, facilitated by dynamic nanoscale interactions, enables structural interconversion in response to external stimuli, providing a unique advantage over previously reported chiral materials.
Beyond chiral self-assembly, this work explores the role of perfluorinated copolymer assemblies in energy storage, environmental remediation, and biomedical imaging. Solid polymer electrolytes based on these assemblies enhance the performance of sodium metal batteries, while fluorophilic interactions in perfluoropolyether-based sorbents enable efficient removal of
perfluoroalkyl substances. Additionally, computational modeling reveals how polymer segment mobility influences fluorine-based MRI contrast agents, optimizing their imaging performance.
Finally, molecular dynamics simulations are employed to study key self-assembly
processes, including hydroxyapatite mineralization in bones and light-driven self-assembly of nanoscale building blocks. By integrating experimental and computational approaches, this thesis advances the fundamental understanding of self-assembly mechanisms and offers new strategies for designing functional materials with broad applications in sensing, separation, biomedicine, and sustainable technologies
Optimizing Key-Value Store Systems Across Multiple Dimensions
No full textKey-value stores play a crucial role in modern data management applications, supporting a wide array of applications like cloud services and real-time data analytics. Despite extensive study and optimizations over decades, these systems still face challenges posed by fast-evolving hardware design and application needs, and careful new designs are necessary to meet these requirements. This dissertation examines three key dimensions in key-value store systems—storage management, transaction processing, and concurrency control. Targeting these dimensions, it introduces novel optimization techniques that strengthen the fundamentals of key-value store systems, aligning them more effectively with modern hardware and application demands
Evaluation of a Wearable Respiratory Protection Assessment System
No full textAnnual fit testing confirms a respirator provides adequate protection to a worker at the time of testing, but does not guarantee stable fit during point of use in the workplace. Current instruments, limited by their size, weight, rudimentary data logging, and reliance on external power, cannot assess fit during actual use. This study introduces the Wearable Respiratory Protection Assessment System (WRPAS) by TSI, Incorporated, designed to provide real-time, in-situ respirator fit measurements. Our objectives were to evaluate the WRPAS's ability to detect respirator leakage, identify which activities most accurately predict overall simulated workplace protection factor (SWPF) fit, and examine how the respirator fit changes over time. The SWPF study was conducted with 19 participants performing 15 tasks of varying intensities, repeated twice. Participants were divided into groups with and without intentional leaks. The leaks were created using a Y-valve that allowed 25% of ambient air particles into the mask. Using Welch’s t-test, significant differences in both activity-specific and overall SWPFs were observed between groups (p < 0.05). Spearman’s Rank Correlation identified five of the fifteen activities as strong predictors of fit: bending over, p = 0.01; facial expressions, p = 0.02; loading and unloading a cart, p = 0.04; army crawl, p = 0.01; and assessing a manikin, p = 0.04. The current study confirms the WRPAS's ability to assess respirator fit changes in real time and under physical exertion, enabling its use in future workplace protection factor assessments
Search for Associated Production of Higgs and V Bosons with Boosted Jet Tagging in Semileptonic Events
No full textThis thesis presents the first search for associated production of a boosted Higgs boson decaying to WW along with a V boson, i.e., a W or Z boson decaying hadronically. The search is performed using data collected with the CMS detector corresponding to 139 fb^-1 of LHC proton-proton collision data at sqrt(s) = 13 TeV. We target semileptonic decays of the Higgs boson, where the hadronic and leptonic W decays are merged inside a wide-radius AK8 jet. Events are selected using machine-learning taggers to identify the Higgs boson jet and the V boson jet
Data sources:
Access Repository Dashboard
Do you manage Open Research Online? Become a CORE Member to access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard! 👇