19304 research outputs found
Sort by
Characterizing and Modeling of Curing, Compressive, Piezoresistivity, and Strength of Smart Cement with Recycled Glass and Plastic Powders
No full textWith rising environmental concerns, there is a growing focus on integrating recycled materials into cement and concrete for sustainable construction. This study investigated the effects of incorporating commercially available recycled glass powder and recycled plastic powder into smart cement, a material known for its piezoresistive properties and has the potential for use in infrastructure applications such as bridges and oil wells. While smart cement provides valuable real-time structural data, a newly developed sensor was integrated to enhance detection, aiming to improve the efficiency and reliability of smart cement in reducing risks of structural failures during construction and also inservice. This study investigated the addition of up to 5% recycled glass powder, recycled plastic powder, or a mix of both (by weight of cement) on the impedance, rheology, curing, and compressive piezoresistivity of smart cement. Both testing and modeling were done the properties such as flow behavior, curing, shrinkage, setting times, and compressive piezoresistive characteristics. Electrical impedance analysis identified resistivity as a critical metric for monitoring. Adding 0.01% carbon fiber notably improved the piezoresistive response, enhancing the material’s sensing abilities. Curing behavior was monitored using the resistivity changes, and the Vipulanandan curing model was used to predict the behaviar. Rheological performance was modeled using the Herschel-Bulkley and Vipulanandan models. The smart cement exhibited significant piezoresistive strain under peak stress, with its response influenced by curing time, water-to-cement ratio, and the type and percentage of recycled content. The stress-piezoresistive strain relationship was accurately predicted using the Vipulanandan p-q model
Faster Optimal Power Flow Using Graph Neural Network-Assisted Methods
No full textThe electrical grid is transforming from a centralized system to a bidirectional network due to the rise of distributed energy resources like residential solar panels. The transformation, coupled with increased demand from electric vehicles and data centers, is putting significant strain on the aging electrical infrastructure. Insufficient investment in modernizing the grid, particularly transmission networks, further exacerbates issues related to the reliability and security of the grid. Independent System Operators manage grid reliability using tools like Optimal Power Flow (OPF) to balance generation and demand efficiently. Increasing grid complexity and congestion makes solving OPF problems quickly a significant challenge. The complexity of the modern grid requires advanced computational methods to solve OPF problems quickly and efficiently. Advanced methods are explored to optimize power system operations by reducing computational complexity in solving OPF problems. These problems are critical for determining generation dispatch points in real-time but become computationally intensive for large power networks due to the numerous variables and constraints involved. Proposed Graph Neural Networks (GNN), a machine learning approach, are utilized to reformulate the traditional OPF problems. The followings are key innovations: • Congested Lines Prediction: GNN predicts heavily loaded or congested lines, allowing the optimization process to focus on critical lines constraints. • N-1 Contingency Management: An augmented hierarchical GNN (AHGNN) is introduced to handle contingency scenarios, resulting in N-1 ROPF models that significantly cut computation time while preserving solution accuracy. • Network Topology Optimization: By integrating GNNs with pre- and post-ML layers, the proposed approach accelerates network-reconfigured OPF (NR-OPF) solutions, improving both efficiency and accuracy. • Maximum-Capacity Generators Prediction: A virtual node-splitting strategy is used to capture generator-level attributes, enabling a two-stage adaptive hierarchical GNN to further reduce constraints and variables, forming the ROPF Lines and Generators model. Case studies confirm that these GNN-accelerated models outperform benchmark models in computational efficiency and solution quality, showcasing the potential of machine learning to enhance real-time grid operation
Lightweight, Post-Quantum Secure Cryptography Based on Ascon: Hardware Implementation in Automotive Applications
No full textWith the rapid growth of connected vehicles and the vulnerability of embedded systems against cyber attacks in an era where quantum computers are becoming a reality, post-quantum cryptography (PQC) is a crucial solution. Yet, by nature, automotive sensors are limited in power, processing capability, memory in implementing secure measures. This study presents a pioneering approach to securing automotive systems against post-quantum threats by integrating the Ascon cipher suite---a lightweight cryptographic protocol---into embedded automotive environments. By combining Ascon with the Controller Area Network (CAN) protocol on an Artix-7 Field Programmable Gate Array (FPGA), we achieve low power consumption while ensuring high performance in post-quantum-resistant cryptographic tasks. The Ascon module is designed to optimize computational efficiency through bitwise Boolean operations and logic gates, avoiding resource-intensive look-up tables and achieving superior processing speed. Our hardware design delivers significant speed improvements of 100 times over software implementations and operates effectively within a 100 MHz clock while demonstrating low resource usage. Furthermore, a custom digital signal processing block supports CAN protocol integration, handling message alignment and synchronization to maintain signal integrity under automotive environmental noise. Our work provides a power-efficient, robust cryptographic solution that prepares automotive systems for quantum-era security challenges, emphasizing lightweight cryptography’s readiness for real-world deployment in automotive industries
Pancreaticobiliary Maljunction and Its Relationship with Biliary Cancer: An Updated and Comprehensive Systematic Review and Meta-Analysis on Behalf of TROGSS—The Robotic Global Surgical Society
No full textObjective: This systematic review and meta-analysis aimed to determine the degree to which pancreaticobiliary maljunction (PBM) increases the risk of different types of biliary cancer (BC). Methods: A systematic review and meta-analysis were carried out using the following databases: PubMed, Embase, Cochrane Library, Scopus, Web of Science, and Science Direct. We systematically searched from inception to April 2024. The search terms included were derived from the keywords “Pancreaticobiliary Maljunction” OR “Anomalous Pancreaticobiliary Junction” AND “Cancer” OR “Malignancy”. Studies that provided data comparing BC rates in relation to PBM presence or vice versa were included. The Newcastle–Ottawa Scale (NOS) was used for quality assessment. The random-effects model was used. Results: Fifteen studies were included with a total sample of 8604 patients, of whom 5015 (58.29%) were female with a mean age of 54.58 years. Patients with PBM had 8.42 (95% CI = 3.57–19.87) more risk of developing any type of BC, with a higher risk of GBC than BDC (OR = 16.91 vs. OR = 3.36, p-value = 0.003). There was a higher risk of having PBM in patients with GBC than BDC only when considering the Asian population (OR = 3.12, 95% CI = 1.09–8.94). Meta-regression analysis revealed that neither mean age (p = 0.087) nor percentage of female patients in the study population (p = 0.197) were statistically associated with the variations in OR for the risk of BC based on the presence of PBM. Conclusions: There is a significant association between PBM and the risk of having BC, mainly GBC when compared to BDC. Most of the studies published reported data from Japanese patients, which limits the generalization of the results. The age of patients and sex were not significantly associated with the relation between PBM and BC. Further prospective studies in broader populations will provide additional details to take measures for screening and early management of PBM and BC
Limit Laws for Non-uniformly Expanding Transformations
No full textIn this dissertation we study the limit behavior of random variables generated by a class of dynamical system. Specifically, we proved distributional convergence to a stable law for non square integrable observables ϕ : [0, 1] → R, mostly of the form ϕ(x) = d(x, x0)^{− 1/α}, on Gibbs-Markov maps. We follow the scheme of the proof given by Tyran-Kaminska in [20, 21], proving the convergence of a counting process to a Poisson process and a form of decay of correlations estimate for a truncation of the observable that ensures the Birkhoff sum of small values do not contribute too much. Then the stable law is obtained in our settings. Based on [17], the result can be extended to more general models as long as they admit a first return time induced map which falls in the category of one-dimensional Gibbs-Markov maps, for instance, intermittent maps
Fundamental Mechanisms of Membrane Immunoassays and Application to Rapid Cancer Diagnosis
No full textLateral flow assays are point-of-care analytical platforms that fulfill the demand for rapid, low-cost, and accessible testing platforms for many different applications including infection diagnostics. However, these platforms struggle with lower sensitivity compared to more elaborate diagnostic tests. We tackled this challenge by introducing a chemiluminescence-based lateral flow assay for highly sensitive detection of COVID-19 in nasal swab clinical matrices. We used HRP and antibody-coupled bacteriophage as LFA reporter particles and an ECL substrate for generating a chemiluminescence signal corresponding to HRP concentration. We were able to achieve a low limit of detection of 25 pg mL−1 SARS-CoV-2 nucleoprotein spike in a nasal swab extract matrix using a FluorChem system to detect the signals, and 100 pg mL−1 when integrated with a POC in-house developed smartphone-based reader. Moreover, we demonstrated that the smartphone-based phage-LFA can statistically distinguish between the LFA signal from COVID-19 positive (N = 15) and negative (N = 11) nasal swab samples (p < 0.001). The proposed smartphone-readable phage-LFA technology allows rapid, affordable, and sensitive detection of COVID-19 and is translatable for other applications. Despite cancer diagnosis typically not being considered urgent, acute promyelocytic leukemia is a type of cancer that requires urgent diagnosis because it can lead to death within a few days due to severe hemorrhagic complications. APL is highly treatable and its early detection can enable successful treatment and prevent early death. We responded to the current demand for timely diagnosis of rapidly fatal acute promyelocytic leukemia (APL) by introducing a sensitive time-resolved fluorescent-based LFA to detect the hallmark of APL, the PML−RARα fusion protein, in cell lysate. Our LFA achieved sensitivity sufficient to meet the WHO guideline of detecting at least 20% blasts. We were able to detect at least 10% NB4 cells (promyelocytic blasts) in healthy PBMC and 20% in HL60 AML cell line background. We also showed the specificity of this LFA by comparing the signal from NB4 APL cell lysate to signals from multiple non-APL cell lines. Finally, we introduced the use of rapid filtration to allow measurement of the adsorption of biomolecules in porous matrices on the short time scales associated with lateral flow assays and membrane chromatography, below one second. We studied the adsorption of fluor-dyed IgG molecules in solution to protein A immobilized on a porous membrane. We showed this technology can be utilized to assess the adsorption under various conditions at timescales not achievable by other methods
Multimodal Characterization of Motor Impairment After Stroke Using Rehabilitation Robotics
No full textA multimodal approach, combining brain, muscle, kinetic, and kinematic data, is increasingly utilized to quantify post-stroke motor impairment in both cross-sectional and longitudinal studies. As the limitations of current clinical scales become more evident, integrating neuro-musculoskeletal data and analyses is emerging as a promising direction for more precise evaluation of rehabilitation outcomes. This dissertation examined motor deficits following a stroke by utilizing innovative assessment techniques and analyzing muscle and joint functions. The study found that alterations in the preferred activation direction of specific muscles contribute to motor impairments by disrupting muscle synergies and impacting isometric force generation. The reliability of an exoskeleton-based system for capturing joint kinematics was established, showing minimal interference with natural movement and a strong correlation with optical motion capture systems. Furthermore, a newly developed standardized objective assessment method demonstrated that the developed score showed a significant correlation with clinical scores. It provided both a comprehensive overview and detailed insights into motor function. Analysis of kinematic synergies uncovered patterns of merging, preservation, and absence in stroke kinematic synergies from healthy kinematic synergies. It offered an intuitive understanding of joint coordination after a stroke
Single-shot X-ray Phase Contrast and Dark-field Imaging with a Single-mask Set-up
No full textTraditional X-ray imaging, which relies on attenuation contrast, faces challenges in differentiating materials with similar attenuation coefficients, particularly in soft tissues. X-ray phase contrast imaging (XPCI) and dark-field (DF) imaging address this limitation by detecting phase shifts and ultra-small-angle X-ray scattering (USAXS), providing enhanced contrast. However, existing XPCI and DF methods often require multiple exposures along with complex and costly setups, hindering their widespread adoption. In this study, we introduce a novel single-mask X-ray phase contrast and dark-field imaging and micro-CT setup capable of simultaneously capturing attenuation, differential phase contrast (DPC), and dark-field images in a single exposure. Most importantly, our proposed design allows effectively capturing multiple contrast features using mask alignment with relatively low resolution detectors. We propose three variations of the single-mask setup, each optimized for different contrast modalities, offering flexibility and efficiency in a variety of applications. A novel physics model for our imaging system based on the x-ray Fokker-Planck equation is presented, which provides an intuitive understanding of signal and contrast formation in single-mask x-ray phase imaging, offering a clear perspective on the image formation process, as well as retrieval method for each contrast types. To further support system design and optimization, we propose a diffraction-corrected Monte Carlo simulation method that combines ray and wave optics models, enhancing our ability to study and refine the imaging system. Our approach eliminates the need for highly coherent X-ray sources, ultra-high-resolution detectors, or intricate gratings. This significantly simplifies the imaging process thereby improving the prospects of clinical translation with a simple and low-cost setup. The versatility of this single-mask approach holds promise for broader use in clinical diagnostics and industrial inspection, making advanced X-ray imaging more accessible and cost-effective
Predictive Analytics for Electronic Health Records (EHR) Data with Deep Learning
No full textPredictive analytics using Electronic Health Records (EHR) has grown considerably in recent decades due to the emergence and adoption of new methods, such as Deep Learning and Artificial Intelligence, to face several diagnostic challenges. During the COVID-19 pandemic, many of these diagnostic challenges have appeared, including the post-infectious sequelae multisystem inflammatory syndrome in children (MIS-C). This syndrome shares several clinical features with other entities, such as Kawasaki disease (KD) and endemic typhus, among other febrile diseases. Endemic typhus, or murine typhus, is an acute infection treated much differently than MIS-C and KD. Early diagnosis and appropriate treatment are crucial to a favorable outcome for patients with these disorders. To address these challenges, different AI-based algorithms using EHRs can be implemented to support medical teams' decision-making to differentiate between these febrile conditions. This research proposes two Clinical Decision Support Systems (CDSSs), one based on an Attention-LSTM network, AI-MET, and another based on a Triplet Loss Siamese network, AI-HEAT, to distinguish between patients presenting similar pediatric febrile conditions using clinical and laboratory features typically available within six hours of presentation. The study utilizes a comprehensive dataset of pediatric patients diagnosed with MIS-C, KD, or endemic typhus, collected from multiple healthcare institutions across the United States. The findings presented in this work are strong evidence that the two robust AI-based CDSSs proposed can accurately differentiate between MIS-C, KD, and endemic typhus using early clinical and laboratory data available during the first six hours after the patient's arrival. These tools have the potential to significantly improve the diagnostic process for clinically similar pediatric febrile conditions, ultimately leading to more timely and appropriate treatment decisions
Agentic Framework for Domain-specific RAG Evaluation
No full textLarge Language Models (LLMs) have revolutionized generation and understanding of textual information and have wide-ranging applications. While the capabilities of LLMs are immediately impressive to any user, extended use quickly reveals problems of inaccurate text generation associated with these models. Retrieval-Augmented Generation (RAG) approaches are used to enhance LLM reliability by grounding responses in knowledge bases, significantly reducing hallucinations. However, RAG performance is highly domain sensitive, necessitating careful tuning of components before deployment for specialized applications. This challenge underscores the critical need for robust evaluation frameworks tailored to domain-specific RAG systems. Existing methods often rely on heuristic-based metrics such as exact match or BLEU scores, which fail to capture deeper semantic reasoning and nuanced understanding. Additionally, these frameworks typically depend on manually curated Question-Answer (QA) datasets, which are often unavailable or insufficient in specialized domains. To address these limitations, we propose an Agentic Framework for Domain-Specific RAG Evaluation. Our approach introduces a synthetic data generation pipeline that simplifies the adaptation of RAG systems to new domains. We incorporate LLM-as-a-judge metrics to enable a more holistic and versatile evaluation of both synthetic datasets and RAG performance. We present MiliQA, a synthetic data set derived from military documents, and compare its quality against public data sets of QA such as Aurelio Mixtral, HuggingFace QA, and WikiEval. To validate our metrics, we benchmark them against human-annotated datasets, including STS-B and SQuAD 2.0. Finally, we demonstrate the applicability of our framework by evaluating key components of RAG systems including embedding models, LLMs, and multiple RAG methodologies, using MiliQA. The results demonstrate the practical value of our proposed framework for guiding the design and optimization of RAG systems for domain-specific applications