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
Temporal Reasoning in Clinical Narratives: From Information Extraction to Early Disease Detection
No full textTimely and accurate prediction of chronic disease risk, such as Type 2 Diabetes, requires models that can reason over complex, longitudinal clinical narratives. This thesis presents a unified framework for temporally grounded patient modeling that integrates structured event representations, fine-grained temporal reasoning, and scalable predictive architectures.
Initial experiments show that concept-based models paired with visit-level temporal modeling outperform traditional baselines that ignore temporal dynamics, motivating a deeper exploration into concept and temporal relation extraction from clinical narratives. Towards this, I introduce GraphTREx, a state-of-the-art temporal relation extraction approach that achieves a 5% F1 improvement on the end-to-end temporal relation extraction task in the I2B2 2012 challenge with ~9% gains on long-distance relations. GraphTREx generalizes well, showing strong performance on the E3C corpus and robust out-of-domain results (without additional adaptation) on a newly annotated UI Health dataset with dense temporal relations, highlighting its portability.
Building on these temporal graphs, I develop HiTGNN—a Hierarchical Temporal Graph Neural Network that integrates intra-document temporal relations, inter-visit dynamics, and external medical knowledge, enabling reasoning across both local event structures and longitudinal patient trajectories. I also introduce ReVeAL (Reasoning with Verifier Aided Labeling), an inference-time scaling framework where a smaller LLM validates predictions from a larger frozen LLM, inheriting interpretability and improving accuracy without full retraining. While our LLM-based approach offers explanations, HiTGNN outperforms it (and other LLM baselines) in accuracy and efficiency, especially over shorter prediction horizons, demonstrating the strength of lightweight, temporally grounded graph-based models, well-suited for low-resource, privacy-sensitive clinical settings.
Rigorous dataset curation, ablations, and analyses reinforce robustness of these approaches. Together, these contributions establish a generalizable framework for clinical prediction by integrating concept abstraction, temporal structure, and semantic enrichment to support robust, privacy-conscious AI systems for real-world clinical decision support
Spatiotemporal Dynamics of Snow-Free Frost Exposure in the Western Cordillera Seasonal Snow Zones
No full textIn the mountain ecosystems of the Western Cordillera, seasonal snowpack provides insulation for soils and vegetation against freezing temperatures. While temperatures are warming overall, concurrent declines in snowpack are removing this protective thermal blanket. This may lead to events where the ground surface becomes more vulnerable to frost damage during synoptic-scale cold events, such as arctic air incursions, even as average temperatures rise. The net impact of these competing effects on ground-level frost exposure remains poorly understood. This thesis addresses this gap by analyzing the spatiotemporal dynamics of snow-free frost days (SFFDs), defined as days with maximum air temperatures below 0°C and minimal snow cover (≤20 mm Snow Water Equivalent), across the Western Cordillera. Using a quality-controlled dataset of daily temperature and snowpack records from 414 Snow Telemetry (SNOTEL) sites over a 31-year period (1990-2020), this study quantifies long-term trends in SFFD frequency, identifies the primary drivers of these trends, and assesses ecological impacts using a satellite-derived vegetation index (NDVI). The analysis reveals that SFFDs are a relatively rare phenomenon with occurrences concentrated in the fall when the snowpack is still developing and its insulating capacity is inconsistent. Results show a statistically significant network-wide decrease in SFFD frequency over the study period. This decline is mainly driven by increasing temperatures that have thus far outweighed the impact of a less persistent snowpack at most sites. The analysis suggests that grassland sites in areas with historically reliable snow cover (low-frost exposure) maintain higher productivity relative to sites with intermittent snow cover (high-frost exposure), and this gap widens during colder years, highlighting the importance of the snowpack’s insulating capacity on this ecosystem. This analysis did not find evidence that conifer ecosystems were affected by changes in SFFDs
β-Uniform Convexity and Divisible Domains
No full textDivisible convex sets have long been important in the study of Hilbert geometries. When a divisible convex set is an ellipsoid, the Hilbert geometry it induces is the hyperbolic space. In general, strictly convex divisible domains exhibit negative curvature properties, but only the ellipsoid is a CAT(-1) space. The notion of p-uniform convexity from the theory of Banach spaces has been proposed as a generalization of the Alexandrov-Toponogov comparison property
to Finsler manifolds. We prove that a natural Finsler metric on a strictly convex divisible domain is β-uniformly convex, where the exact constant β is related to the regularity of the boundary
Effect of Overlying Material on EndoCem Mineral Trioxide Aggregate Setting Reaction in Primary Molar Pulp
No full textPurpose: To evaluate the impact of common restorative materials on the microhardness of ENDOCEM-MTA in primary molar pulpotomies. The study assessed whether the material placed over ENDOCEM-MTA affects its setting reaction, surface hardness, and uniformity across depths.
Methods: Fifty extracted primary molars were prepared using standardized pulpotomy protocol. A 3mm layer of ENDOCEM-MTA was applied to the pulpal floor. Specimens were divided into five groups based on the overlying material: control (moistened cotton pellet), resin-modified glass ionomer cement (RMGIC, Fuji II LC), zinc oxide eugenol cement (IRM), RMGIC combined with a stainless-steel crown (SSC), and IRM combined with SSC. Teeth were incubated at 37°C and 100% humidity for 24 hours to simulate oral conditions. After incubation, specimens were sectioned, polished, and analyzed for microhardness using a Vickers Microhardness Tester at depths of 1mm, 2mm, and 3mm. Statistical analyses, including ANOVA and Tukey post hoc tests, were performed with significance set at P < 0.05.
Results: All experimental groups performed as well as or better than the control, demonstrating clinical applicability. Group 4 (RMGIC with SSC) achieved the highest mean Vickers Hardness Number (VHN) of 72.66 ± 6.34 at 3mm depth, while the control group exhibited the lowest VHN (58.42 ± 11.26 at 1mm depth). Microhardness remained consistent across all depths, with significant differences observed between materials (P = .0009).
Conclusions: Restorative materials significantly influenced the microhardness of ENDOCEM-MTA. RMGIC and IRM combined with SSC produced the most favorable outcomes, supporting their use in single-visit pulpotomies for improved procedural efficiency
Development of Electrochemical Immunosensor for Oral Cancer: Initial Outcome
No full textOral Squamous Cell Carcinoma (OSCC), accounting for nearly 90% of head and neck malignancies, continues to pose a global health challenge with over 250,000 new cases annually, especially in developing nations. Its frequent progression from precancerous lesions such as oral leukoplakia (OL) underscores the urgent need for an early, non-invasive diagnostic tool. This thesis investigates the use of a nanoengineered electrochemical immunosensor targeting the S100A7 protein, a biomarker associated with OSCC risk. The biosensor's sensitivity and specificity were assessed across a range of clinically relevant concentrations in artificial serum and artificial saliva, simulating different stages of the disease. Additionally, an ELISA assay was performed to validate sensor performance and benchmark biomarker detection.
The findings demonstrate consistent trends in biomarker expression and validate the potential of the sensor for early-stage diagnosis. This integrated approach highlights a promising pathway towards non-invasive, point-of-care screening for malignant transformation in patients with oral premalignant disorders such as OL. However, further research involving clinical samples and larger datasets is necessary to establish the sensor's diagnostic accuracy and real-world applicability
Creep Damage Evolution in 316L Stainless Steel: Correlating Cavitation with Acoustic Emission
No full textThis study investigates the creep deformation characteristics of conventionally manufactured
316L stainless steel at an elevated temperature of 650°C, assessed under three distinct quasistatic
stress levels (241 MPa, 282 MPa, and 311 MPa). The investigation is particularly focused
on the progression of damage within the diffuse necking region, an antecedent to catastrophic
rupture. A comprehensive post-mortem microstructural analysis, utilizing the precision of scanning
electron microscopy (SEM), was conducted to thoroughly evaluate the morphology and
distribution of creep-induced defects, including pores, grain boundary cracks, and intergranular
voids. In parallel, the application of real-time acoustic emission (AE) sensing was employed
to detect and analyse the elastic wave activity generated by microstructural failure phenomena
and creep stages. Through the adept integration of statistical analysis of microstructural defects
with sophisticated AE signal interpretation, the study elucidates a strong correlation between
the applied stress levels and the trajectory of damage evolution. At the lower stress level (241
MPa), fewer but significantly coarser voids were observed, whereas at the higher stress level (311
MPa), numerous localised microcracks formed with limited coalescence. This dual characterization
approach effectively links dynamic AE signatures to the underlying damage mechanisms,
thereby crucially addressing a significant gap in the real-time assessment of creep degradation in
stainless steels. The resulting findings enhance the predictive capability of material degradation
and support advanced structural health monitoring strategies for high-temperature applications,
particularly within the scope of Generation IV nuclear reactor component
Essays in Applied Microeconomics
No full textExisting literature demonstrates that peer age plays a crucial role in shaping student outcomes in elementary school. In the first chapter, I build on this literature by highlighting that differences in developmental rates between boys and girls likely lead to varying theoretical effects of peer age based on gender. I find that both male and female peer age positively influence student test scores, but male peer age has an impact that is twice as large as female peer age. Further analysis reveals substantial differences in these effects based on students’ gender, age, and economic disadvantage. Referring to the differential patterns of brain development and hormonal regulation by gender, the second chapter examines the potential differential impact of age on academic performance across genders and subject domains. I find that across both math and reading, girls consistently show higher age-related gains than boys, suggesting faster maturity growth with age
Control System for Outdoor Use of Powered Portable Ankle-Foot Orthosis
No full textTechnological advancements and the decreasing cost of sensors and controllers have fueled a growing interest in gait analysis and support systems, especially for use outside of a laboratory environment. This thesis presents a control system for a lower-limb exoskeleton, the Powered Portable Ankle-Foot Orthosis (PPAFO), designed to assist both individuals with motor disabilities and healthy subjects in various outdoor settings.
The work begins by reviewing relevant literature and prior work in the field of gait analysis, embedded systems, and powered exoskeletons. It then details the implementation of the control system, outlining both the hardware components, such as IMUs, encoders, and industrial computers, and the software components, including the control algorithms and communication protocols. A central feature of the system is the Zero Velocity Update (ZUPT) method, which is used to correct for sensor drift and accurately detect when the foot is stationary during the gait cycle. This allows for the precise, real-time estimation of foot orientation and tibia angle.
The second part of the thesis presents a numerical evaluation of the control algorithms through simulation studies and self-collected data. The performance and adaptability of the system are assessed by comparing the output of a Look-Up Table (LUT) against gait variations in simulated subjects. The results show that while a general LUT offers some compatibility, performance significantly improves with condition-specific LUTs that incorporate factors like terrain inclination. A sensitivity analysis on the ZUPT algorithm and IMU characteristics further demonstrates the system's robustness to noise and varying parameters. The findings also reveal that the angular velocity of the foot can substitute for the tibia's angular velocity at critical moments, suggesting a potential for sensor reduction. Finally, the thesis provides recommendations for future work to enhance the PPAFO's capabilities and robustness in real-world conditions
Association Between Malocclusion and Oral Health Related Quality of Life in Mixed Dentition
No full textAssessing OHRQOL and Interceptive Orthodontics In The Pediatric Population Narayan SL1, Allareddy V2, Avenetti DM1, Hill BJ1,Desai BK1 1Department of Pediatric Dentistry, UIC College of Dentistry, Chicago, IL, 2Department of Orthodontics, UIC College of Dentistry, Chicago, IL Hypothesis and Objective: The objectives of this study was to understand how the OHRQOL in the pediatric population was impacted prior to being treated with interceptive orthodontic appliances at the University of Illinois Chicago College of Dentistry (UIC COD). Methods: To conduct this study, a questionnaire was distributed to the legal guardians and children of six to eleven years attending UIC COD for their dental examinations. Data regarding the child’s orthodontic status and tentative treatment plan were extracted from their electronic dental records. Results: Forty eight children (Median age of 8) and parents were recruited to participate in the study. Thirty- nine percent of children were male and sixty percent were female. Fifty- six percent of the children identified as White, 30% as Black or African American, 5% as Asian, and 7% as American Indian. Twenty- six percent presented with posterior crossbite, 46% anterior crossbite, 44% with crowding, and 14% with either space loss, anterior open bite, or deep bite. Conclusions: Research has shown that interceptive orthodontics has had a positive impact on a patients OHRQOL . Further studies regarding the intervention of interceptive orthodontics can accelerate the understanding of OHRQoL
Raw data for <b>Evaluation of BR1 and BI30 AAVs for brain endothelial tropism</b>
No full textRaw data for publication Evaluation of BR1 and BI30 AAVs for brain endothelial tropism</p
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! 👇