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
Accelerating Material Inverse Design in High-Dimensional Continuous Spaces with Autonomous Machine Learning Ecosystems
No full textModeling and designing materials at the nanoscale is essential for advancing new technologies in crucial areas such as energy, electronics, and catalysis. By gaining control over structure at the atomic level, we can now pursue inverse design, where the goal is to start from a desired property and determine the atomic structures of materials that can achieve it. However, this process is far from straightforward. The design space is vast, complex, and often difficult to explore. Traditional physics-based methods such as Density Functional Theory (DFT) and Molecular Dynamics (MD) continue to provide high accuracy in exploring the material design space at the nanoscale, but they are typically too slow or computationally expensive when applied to the enormous range of possible material compositions and configurations. In recent years, machine learning (ML) and artificial intelligence (AI) have emerged as promising tools. These techniques can create fast, predictive models and generate new design strategies, helping reduce computational cost while still delivering accurate and robust predictions. Despite these advances, building a fully autonomous system for materials discovery remains a major challenge. It requires the integration of multiple steps, including conceiving a material idea and searching for optimal candidates, gathering quality data for efficient learning, learning meaningful physical patterns, constructing reliable models, evaluating phase stability, and identifying viable synthesis routes, all within a coherent and intelligent workflow.
This thesis takes on these challenges by creating and applying machine learning frameworks across major computational materials research themes crucial for building an effective and sustainable materials ecosystem. First, it explores smart ways to sample large design spaces of materials using reinforcement learning. This is applied to both discrete problems, such as optimizing defects in 2D materials, and continuous problems, such as predicting crystal structures to inverse design superhard materials or tuning mechanical parts as a continuum-scale modeling problem. Tests on standard benchmark problems show that the proposed approach often performs better and faster, especially in complex non-convex search spaces with many local optima. Second, the thesis introduces learning methods for generating representations of atomic environments. These representations are a key necessity for a machine learning or AI model to effectively learn the physics behind any physical phenomenon in materials. A graph-based neural network architecture is introduced that learns symmetry-aware, noise-insusceptible, unique, and robust features for material characterization. These features help us better understand phase changes and link structure to properties in materials such as porous frameworks like zeolites, nanoparticles, and supercritical fluids. Third, the thesis delves into the development of foundational models and their testing, particularly through the use of machine learning-based interatomic potentials, including Gaussian Approximation Potentials (GAP) for elemental nanoclusters and more advanced equivariant models such as MACE for 2D transition metal dichalcogenides. These models can predict energy, forces, and structural behavior very accurately, offering faster and more scalable alternatives to traditional physics-based approaches. These models can serve as accurate and computationally efficient predictors for materials properties across a wide thermodynamic window, in an efficient and scalable manner. Fourth and finally, the thesis explores the application of these interatomic models to understand and implement molecular simulations for studying complex phase behaviors and dynamics at the nanoscale—such as metastable phase formation in amorphous ice, low-friction (superlubric) transitions, changes in grain boundaries in ice films, and metal–insulator transitions in certain oxides. Using symbolic regression and data-driven techniques, the thesis extracts simple and interpretable models that explain and predict these complex behaviors.
Building on these capabilities, the final section of this thesis envisions autonomous materials design and discovery ecosystem. At its core is the integration of large language model (LLM) as AI agents serving as cognitive orchestrators—able to parse literature, generate simulation workflows, reason across symbolic and numerical domains, and iteratively improve based on experimental or computational feedback. These agents go beyond automation to function as scientific collaborators, supporting hypothesis generation, synthesis planning, and cross-domain optimization. Proposed applications include explainable phase prediction in zeolites, optimization of high-entropy alloys, and reinforcement learning-based recipe generation for autonomous synthesis. By interfacing with simulations, databases, and experimental data, these agents can enable real-time, closed-loop materials discovery
Longitudinal Alignment of ACGME Milestones for General Surgery Residency and Vascular Surgery Fellowship
No full textObjective: To understand the residency to subspecialty fellowship transition by exploring the alignment of graduating general surgery and initial vascular surgery fellowship ACGME Milestones 2.0 within: (1) harmonized and (2) non-harmonized competencies.
Background: The harmonization of Milestones 2.0 across specialties presents an opportunity to evaluate a trainee’s progression from residency through fellowship. Understanding the alignment between general surgery residency and vascular surgery fellowship is important to identify struggling learners earlier and explore this transition.
Methods: This is a retrospective review of a national cohort of vascular surgery fellows (starting 2022 and 2023) who previously completed general surgery residency. We compared ACGME Milestones from the final 2 years of residency to the first year of fellowship using descriptive statistics and a mixed effects regression model.
Results: Milestones data were collected for 266 vascular surgery fellows from 2021-2024. PGY-5 end-year compared to year-1 vascular surgery fellowship end-year Milestones found a coefficient of 0.20 for PC (P=0.004), 0.13 for MK (P=0.027), 0.23 for PROF (P<0.001), and 0.17 for ICS (P=0.01). SBP and PBLI were not significantly aligned for PGY-5 end-year. The matched subcompetencies showed similar significant alignments: 3 of 4 PC, 2/2 MK, 4/4 PROF, and 2/3 of ICS subcompetencies.
Conclusions: There is significant alignment between graduating general surgery and year-1 vascular surgery Milestones for non-harmonized competencies, PC and MK, and 2 of 4 harmonized competencies: PROF and ICS. These findings add internal structure validity evidence to Milestones 2.0 with potential to use the residency-fellowship hand-off to institute early improvement efforts for trainees
A Frequency Loss Driven Framework for Respiration Monitoring Using Depth-Sensing Cameras
No full textRespiratory monitoring is crucial for detecting physiological distress, yet traditional contact-based methods are often intrusive and impractical for long-term use. Non-contact approaches, particularly camera-based techniques, offer a promising alternative, but conventional RGB-based systems raise concerns regarding data complexity and privacy. Depth-sensing cameras,
by contrast, inherently safeguard privacy and simplify data handling, while remaining robust to lighting variations. However, challenges such as subtle respiratory motion, occlusions, and noise continue to reduce the effectiveness of depth-based respiration monitoring. To address these limitations, this research presents a deep learning framework tailored for breathing signal
extraction and respiratory rate prediction solely from depth images. The proposed architecture combines a Convolutional Neural Network (CNN) for spatial feature extraction with a Long Short-Term Memory (LSTM) network to model temporal dynamics. A novel frequency domain loss function further guides the training process, encouraging accurate estimation of the dominant respiratory frequencies. The model was trained and evaluated using the pub
licly available ”Breathing In-Depth” dataset, encompassing diverse breathing rates, postures, and subject-camera distances. Experimental results demonstrate that the frequency-optimized model significantly outperforms traditional time-domain training approaches, achieving superior respiratory rate accuracy while maintaining resilience against noise and occlusions. These
findings highlight the potential of depth-only, frequency-driven frameworks for robust, privacy-preserving respiratory monitoring
Innovative Systemic Control of Grid-Edge Inverters in Distribution Grids for Enhanced Grid Services
No full textDistributed Energy Resources (DERs), particularly inverter-based energy resources, are critical for achieving decarbonized power systems. However, their stable integration into weak distribution networks presents significant challenges. To address these challenges, this thesis develops a comprehensive, multi-layered control framework that enhances the performance and resilience of grid-edge inverters. Specifically, it proposes an innovative systematic methodology to improve the grid services of grid-forming inverters, ensuring robust performance under complex grid conditions. This methodology demonstrates broad applicability across various grid-forming technologies and renewable energy plants, highlighting its scalability. Furthermore, a novel coordination strategy is developed to enable effective operation of inverter-based energy resources. To further strengthen system adaptability, AI-enhanced methods are incorporated to improve system resilience. These methods are validated to effectively improve the overall resilience and performance of modern distribution grids
SOX2 is Associated with Changes in DNA Damage Repair and Cell Cycle Gene Expression and Regulates E2F1
No full textA leading cause of disease progression in prostate cancer patients is therapy resistance. As a result, different treatment schemes and combinations are employed to combat prostate cancer, advancing to more deadly prostate cancer. One mechanism that cancer cells utilize to evade treatment is through alterations in DNA damage repair pathways. Currently, the tumor expression profiles score for homologous recombination deficiency is available for patients. However, this is typically only used to determine if Poly (ADP-Ribose) Polymerase (PARP) inhibitors are an appropriate treatment option.
One potential biomarker my lab has identified is SOX2. The expression of SOX2 was linked to an increased Gleason score at diagnosis and decreased time to metastasis after biochemical recurrence. Previously, SOX2 was found to promote lineage plasticity and resistance to AR-targeted therapies in prostate cancer. Further, the SOX2 protein is either present or absent in a prostate cancer tumor, making it an ideal biomarker.
Our studies show that metastasis samples from prostate cancer patients with high SOX2 mRNA expression have gene enrichment in cell cycle and DNA damage repair pathways. One protein that regulates cell cycle progression and transcribes DNA damage repair proteins is E2F1. This thesis found that E2F family gene targets are upregulated in response to increased SOX2 expression and that SOX2 targets the E2F1 promoter. Currently, there are few biomarkers for treatment outcomes in prostate cancer. To address this gap, I aim to better understand SOX2 in prostate cancer disease to build a foundation for SOX2 as a predictive marker in patient care
A Computational Approach to Efficient Electrode Material Design for Electrochemical Oxidation of PFAS
No full textPer- and polyfluoroalkyl substances (PFAS) are persistent contaminants of growing environmental concern due to their strong carbon-fluorine bonds and resistance to conventional degradation methods. Among destructive treatment technologies, electrochemical advanced oxidation processes (EAOP) have shown great promise for PFAS remediation in water. However, the reaction mechanisms are inadequately understood, which hinders the development of design principles for efficient electrode materials.
This thesis presents a computational framework to advance the understanding of the oxidation mechanism of EAOP, focusing on perfluorooctanoic acid (PFOA) as a model contaminant. Through quantum mechanical simulations, the initial steps of EAOP, mainly the decarboxylation and chain-shortening reactions, are systematically explored. The effects of surface termination, PFOA orientation, explicit solvation, and applied potential are first evaluated on the Ti4O7 [112] surface, revealing that decarboxylation is favored for specific orientation of PFOA, and takes place under applied potential. Energetic barriers for decarboxylation were calculated next. These insights are then extended to other electrode materials such as boron-doped diamond (BDD), SnO2, Bi2O3, and RuO2, highlighting specific influences on the chain-shortening reactions. Additionally, it is shown that the major competing reaction for EAOP is the oxygen evolution reaction (OER). Prevention of OER is thus key to designing efficient electrode materials for EAOP. Building on these mechanistic insights, the thesis transitions toward accelerating discovery of materials that can prevent OER. For this purpose, graph neural network (GNN) models are shown to be an efficient pathway to screen a diverse library of electrode materials by accurately predicting the parameters involved in OER.
Overall, this work bridges atomistic insight and data-driven modeling to establish a theory-guided approach for electrocatalyst design. It contributes a fundamental understanding of the mechanism of EAOP, while laying the groundwork for rapid discovery of advanced materials for PFAS remediation
Regenerative Electric Spring Based Grid Forming Inverter for Next-Generation Power Systems
No full textWith the transition towards renewable energy sources (RESs), traditional power systems are evolving into power electronics-dominated grids. The reduction of synchronous generator-based inertia increases grid vulnerability to frequency and voltage fluctuations, requiring advanced control strategies to enhance stable and reliable operation under varying load and generation conditions. In addition, the growing adoption of electric vehicles (EVs) and battery energy storage system (BESS) presents new opportunities for grid support. By leveraging Grid forming inverter (GFMI) control, these technologies can provide supplementary services that enhance grid resilience and flexibility.
Unlike traditional GFMI control strategies, this thesis explores using electric spring-based GFMI with BESS to enhance frequency response and support the grid during peak demand periods. This approach enables controlled power injection when needed while absorbing excess power during off-peak times, helping maintain grid balance. By leveraging the advantageous properties of mechanical springs—such as robustness, adaptability, and fast response time—this method ensures rapid intervention, efficient charging and discharging, and improved coordination with other inverters in the system.
Building on this foundation, the research further explores the role of grid-forming inverters in integrating vehicle-to-grid (V2G) technology, where EVs contribute to grid support through frequency and voltage regulation and peak shaving. A control strategy is proposed to optimize power injection from EVs based on various factors, ensuring efficient energy exchange between EV charging stations and the grid while taking advantage of the storage capabilities of EV batteries.
Several control strategies have been proposed in the literature to enhance frequency response and stability in power electronics-dominated grids. In (1), a supercapacitor-based energy storage system enhances grid stability and resilience. In (2), a flywheel storage system is used for grid support and peak shaving applications. In (3), worn-out batteries support the grid with an algorithm that optimally manages each battery. In (4), an additional grid-forming inverter was applied to critical loads to prevent frequency and voltage fluctuations.
This thesis presents a supervisory control framework integrating control algorithms for BESS and EVs interacting with GFMI to support the grid. In both cases, simulations were conducted in MATLAB/Simulink to evaluate the effectiveness of the proposed control strategies. The controls dynamically adjust power injection based on real-time grid conditions, ensuring stability, fast response, and efficient energy utilization.
Simulation results demonstrate improved frequency stability, reduced battery stress, and enhanced power-sharing efficiency, validating the effectiveness of the proposed approach. This framework enhances the performance of power electronics-dominated systems by enabling a more adaptive and resilient grid support mechanism
Reinforcement Learning for Multi-modal Human-Robot Interaction
No full textThis dissertation addresses the challenges of enabling natural and effective collaboration between humans and assistive robots in domestic settings. To support users in activities of daily living (ADLs), assistive robots must understand multimodal inputs, engage in dynamic interaction, and adapt to vague or evolving instructions. We approach this problem through a unified framework that spans low-level interaction policy learning, user simulation, and high-level task planning.
First, we propose a neural network-based multimodal user simulator trained on real-world demonstrations from the ELDERLY-AT-HOME corpus. The simulator generates realistic human behavior across speech, gestures, and haptic actions, enabling scalable reinforcement learning (RL) for collaborative tasks. We then develop an interpretable RL-based Interaction Manager that learns to take the role of a helper (HEL) in a collaborative object-finding task, demonstrating high accuracy and user satisfaction through a comprehensive user study.
Finally, we introduce IteraPlan, an iterative task-level planning framework that uses Large Language Models (LLMs) to generate and refine plans from vague natural language commands. Unlike existing systems that rely on structured prompts and fixed plans, IteraPlan integrates real-time human feedback to adapt its behavior on the fly
Comparative Efficacy of MUS System vs Er:YAG Laser System in Reducing Bacterial Biofilms in Root Canals
No full textThis in vitro study aimed to evaluate the efficacy of biofilm removal using two advanced irrigation systems: the Multisonic Ultracleaning System (MUS) GentleWave® with the CleanFlow™ handpiece, and the erbium-doped yttrium-aluminum-garnet (Er:YAG) laser (LightWalker® Fotona) for Laser-Activated Irrigation (LAI). The modes used in LAI included Photon Induced Photo-acoustic Streaming (PIPS) and Shock Wave Enhanced Emission Photoacoustic Streaming (SWEEPS).
To conduct this evaluation, forty extracted human mandibular molars meeting the inclusion criteria were collected, processed, and sterilized following Institutional Review Board (IRB) approval. The teeth were inoculated with Enterococcus faecalis and cultured for 21 days to develop biofilms. Biofilm formation in the canals was sampled and confirmed with confocal laser
microscopy. After incubation, the samples were randomly divided into four groups (n=10): Standard Needle Irrigation (SNI), MUS system, Er:YAG SWEEPS, and Er:YAG PIPS. The teeth were instrumented to size 25/V06 and irrigated according to the respective treatment protocols.
Bacterial samples were collected pre- and post-instrumentation. Data were analyzed using an analysis of variance (ANOVA) with a significance level of P < .05, followed by Dunnett’s and Tukey's tests.
The results demonstrated that Er:YAG SWEEPS reduced E. faecalis significantly more than SNI and MUS systems (p 0.05).These findings suggest that enhanced irrigation techniques might improve antimicrobial action against E. faecalis biofilms. The Er:YAG SWEEPS proved to be more effective than Er:YAG PIPS and MUS system within the limitations of the current study
Special Education Teachers' Perceptions of Role Dissonance and Parental Expectations
No full textThe chronic shortage of special education teachers continues to impact educational outcomes for students with disabilities across the United States. While extensive research has examined factors contributing to special education teacher attrition, including role conditions and workplace stressors, limited attention has been given to how parent-teacher relationships influence educators' decisions to remain in the profession. This qualitative study explored the perspectives of ten special education teachers in Illinois who have remained in the field for at least five years, examining their experiences with parental involvement and strategies for overcoming role dissonance.
Using interpretive grounded theory methodology, the study conducted two rounds of semi-structured interviews with participants ranging from 6 to 27 years of teaching experience. The research was guided by three questions: (1) How do special education teachers describe parental involvement? (2) Does parental involvement influence a special educator's long career in the classroom? (3) What strategies and approaches are utilized to overcome obstacles that special educators face in the classroom, specifically to role dissonance, including parental expectations of their role?
Findings revealed that while participants did not explicitly connect parental involvement to retention decisions, deeper analysis showed that meaningful family partnerships significantly impact teacher self-efficacy, job satisfaction, and professional effectiveness. Teachers valued collaborative relationships with parents, particularly those with strong advocacy skills, and reported that these partnerships positively influenced their instructional practices. The study identified role dissonance as an ongoing challenge, with teachers experiencing conflicting expectations from parents, peers, and administrators regarding their professional responsibilities.
Participants employed several strategies to overcome obstacles and remain in the field, including leveraging technology to reduce administrative burden, seeking mentorship throughout their careers, developing empathy-based communication skills, and building trust-based partnerships with families. The research highlights the need for enhanced professional development focused on family engagement, structured mentoring programs, and administrator training in special education law and practices.
This study contributes to understanding special education teacher retention by positioning family-school partnerships as a significant factor in professional sustainability. The findings suggest that fostering collaborative parent-teacher relationships, while addressing systemic issues that create role dissonance, may serve as protective factors against teacher burnout and attrition
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! 👇