The University of Texas at Arlington

Current risk assessment models for predicting ischemic stroke (IS) in patients with atrial fibrillation (AF) often fail to account for the effects of medications and the complex interactions between drugs, proteins, and diseases. We developed an interpretable deep learning model, the AF-Biological-IS-Path (ABioSPath), to predict one-year IS risk in AF patients by integrating drug–protein–disease pathways with real-world clinical data. Using a heterogeneous multilayer network, ABioSPath identifies mechanisms of drug actions and the propagation of comorbid diseases. By combining mechanistic pathways with patient-specific characteristics, the model provides individualized IS risk assessments and identifies potential molecular pathways involved. We utilized the electronic health record data from 7859 AF patients, collected between January 2008 and December 2009 across 43 hospitals in Hong Kong. ABioSPath outperformed baseline models in all evaluation metrics, achieving an AUROC of 0.7815 (95% CI: 0.7346–0.8283), a positive predictive value of 0.430, a negative predictive value of 0.870, a sensitivity of 0.500, a specificity of 0.885, an average precision of 0.409, and a Brier score of 0.195. Cohort-level analysis identified key proteins, such as CRP, REN, and PTGS2, within the most common pathways. Individual-level analysis further highlighted the importance of PIK3/Akt and cytokine and chemokine signaling pathways and identified IS risks associated with less-studied drugs like prochlorperazine maleate. ABioSPath offers a robust, data-driven approach for IS risk prediction, requiring only routinely collected clinical data without the need for costly biomarkers. Beyond IS, the model has potential applications in screening risks for other diseases, enhancing patient care, and providing insights for drug development.

Background: International PhD students represent a significant and growing portion of the U.S. doctoral population, particularly in STEM disciplines. Despite their vital contributions to research and innovation, these students face distinct structural and psychosocial challenges that may hinder academic success, mental health, and career outcomes. Methods: We conducted a systematic review of empirical studies published from 2019 to 2025, focusing on international PhD students in U.S. institutions. A convergent synthesis approach was used to integrate quantitative and qualitative data, analyzing outcomes related to academic persistence, psychological well-being, and career transitions. Findings: Thirty-eight studies met inclusion criteria, spanning multiple disciplines and methodological approaches. Key findings revealed high rates of depression, anxiety, and stress; visa-related career uncertainties; delays in degree completion due to unstable funding and ineffective mentorship; strained supervisory relationships; and weak institutional belonging. These challenges were often compounded, reinforcing systemic disadvantages for international doctoral students.

Vulnerability to sleep disturbance following stress (i.e., sleep reactivity) is associated with incident insomnia, though the underlying physiological mechanisms remain unknown. We examined skin temperature stress responsiveness in individuals with high (HSR) versus low (LSR) sleep reactivity. We hypothesized that individuals with HSR would exhibit exaggerated reductions in distal skin temperature during stress. 28 adults with LSR (5M/9F; age: 21±4 years; BMI: 24±4 kg/m ² ) and HSR (5M/9F; age: 22±4 years; BMI: 23±3 kg/m ² ) participated after completing the Ford Insomnia Response to Stress Test (FIRST). Participants wore a water-perfused suit, which was continuously circulated with 34°C water and covered proximal body regions. All participants underwent a Trier Social Stress Test (TSST) which included baseline, speech preparation, speech delivery, and mental arithmetic phases followed by a recovery period. Distal and proximal skin temperature were monitored throughout the testing session. Skin temperature reactivity did not differ between groups. However, a group effect was observed whereby distal skin temperature was reduced by ~2°C in the HSR compared to the LSR group at all time points. Similarly, upper and lower limb distal proximal gradients (DPG) were ~1-2°C lower in the HSR group (i.e., colder extremities). Higher FIRST scores were associated with more negative DPG (r=0.416-0.578). Despite similar reactivity profiles, individuals with HSR exhibit reduced distal skin temperature, the extent of which was proportional to sleep reactivity severity. These findings suggest that differences in distal skin temperature may be a physiological marker of greater vulnerability to stress-related sleep disturbances in individuals with HSR.

Key-value stores (KV stores) anchored on log-structured merge trees (LSM-tree) provide much improved performance under write-intensive workloads but exhibit significant write amplification (WA). The tiering compaction strategy is widely adopted in KV stores to reduce the WA. However, there are three imminent issues in tiering-based KV stores. First, excessive levels in the LSM-tree structure result in high read and write amplification. Second, without key partitioning support, these KV stores increase write tail latency. Relying solely on a hash-based key partitioning scheme is insufficient, as it does not support range queries. Furthermore, a KV store with only a key range-based partitioning scheme overlooks the distribution of keys within the key space.. Third, existing KV stores with level optimization can incur high costs. A common approach to reducing the number of LSM-tree levels is to increase the MemTable size. While this reduces the number of levels, it requires more memory, leading to higher costs. Therefore, it is crucial to balance the trade-off between level optimization and monetary expenses. To address these issues, we propose a tiering-based key-value store, leveraging high-performance hierarchical data management. The key space of our proposed KV store is partitioned into m ultiple partitions based on key ranges, with an LSM- tree in each partition. We refer to this KV store as MTree in this paper. MTree reduces the number of levels in the LSM-Tree structure close to one without increasing the monetary cost. In MTree, a hierarchical structure that combines the log and LSM-tree curtails the number of levels in the LSM-tree by accumulating the data in the log without increasing the cost. With the hierarchical structure in place, we devise a long- and short-term key density distribution-aware partitioning scheme for the key range. This scheme dynamically adjusts the size of partitioning, balancing data in the partition and storing most data in large-sized logs. Then, MTree reduces the number of levels in the LSM-tree, thereby optimizing the read/write amplification. The experimental results show that MTree limits the WA to as low as two. MTree enhances the random write throughput of the state-of-the-art KV stores by up to 6.9 ×.

We provide a new approach to Raman generation in silicon. Thus, we design and fabricate periodic nanophotonic devices incorporating guided-mode resonance to implement and enhance Raman photon generation. We apply one-dimensional gratings located between two distributed Bragg reflectors for feedback and improved efficiency. Operating in a single polarization state, by spectral and angular tuning, two resonance lines exhibit the proper spectral Raman separation for silicon. The efficiency of Raman photon generation can be enhanced by the resonance Q factor of two split resonant modes with proper grating parameter selection. With the pump at 1529 nm, we detect a Raman signal at 1660.4 nm. This wavelength separation corresponds to a Raman shift of 15.527 THz, which is close to the nominal Raman shift in silicon at 15.606 THz. The Raman generation experimental results match well with the theoretical analysis of split resonance modes enabled by the angular tuning technique. These results demonstrate that Raman enhancement using the guided-mode resonance effect is feasible.

Background Pancreatic adenocarcinoma (PAAD) represents a highly fatal form of cancer. The 5‐year survival rate for patients with this disease is only around 10%. A significant hurdle in its management is the absence of characteristic early‐stage symptoms. As a result, a large majority of pancreatic cancer patients are diagnosed when the disease has reached an advanced stage or has metastasized. Consequently, taking measures to suppress the occurrence of metastasis in pancreatic cancer can bring about a substantial improvement in patients' survival rates and overall prognosis. SKIL , known to promote cancer progression, is implicated in cell proliferation, epithelial–mesenchymal transition (EMT), and metastasis, but its specific function in pancreatic cancer remains unclear. Methods We investigated the effects of SKIL on the proliferation, apoptosis, and metastasis of pancreatic cancer cells. Through ChIP‐seq, we identified the SKIL downstream target gene and further explored the mechanism by which SKIL regulates the metastasis of pancreatic cancer cells through functional experiments and Western blot. Results A high level of SKIL expression is associated with an unfavorable prognosis in PAAD; it promotes cell migration and EMT. Through ChIP‐seq analysis, we identified that SKIL inhibits TSPYL2 , a nuclear protein regulating the TGF‐β pathway by binding to the TGFB1 promoter. Further studies carried out by us confirmed that SKIL modulates the TGF‐β pathway via TSPYL2 , facilitating EMT and metastasis in pancreatic cancer cells, independent of Smad4. Conclusions These findings reveal a novel regulatory mechanism involving SKIL , TSPYL2 , and the TGF‐β pathway, offering new therapeutic targets for PAAD.

This study revisits Sam Dragga’s research on ethical decision-making in document design, updating it to reflect contemporary concerns. Our findings indicate that participants today perceive the document design scenarios as significantly more unethical than those in Dragga's original study, with heightened attention to accessibility, cultural sensitivity, and social justice. While Dragga's study emphasized concerns over the consequences of document design choices, our results suggest a shift in focus toward the writer's intent. Participants frequently judged deliberate manipulation as unethical, even in cases where no direct harm was evident. These findings highlight the evolving ethical priorities in technical communication and underscore the need for practitioners and educators to reassess and revise the field's guiding principles to align with contemporary values of inclusivity and social responsibility.

A detailed understanding of reservoirs is crucial for effective hydrocarbon exploration and production. However, despite technological advancements, accurately characterizing complex reservoirs, such as those in the Lower Goru Formation (LGF) of the Sawan Gas Field (SGF), Pakistan, remains challenging due to heterogeneous lithology. Existing studies often lack a comprehensive integration of seismic, petrophysical, and machine learning approaches, creating a gap in complete reservoir characterization. To address this problem, this study integrates seismic interpretation, well-log analysis, and rock physics modeling to enhance reservoir characterization in the LGF. The stratigraphic and structural features of the study area were analyzed using 2-D seismic lines. Critical reservoir zones were delineated based on petrophysical parameters obtained from several logs. A Self-Organizing Maps (SOMs) based unsupervised machine learning technique was utilized to categorize electrofacies. Cross-plots of P-impedance vs. Vp/Vs and lambda-rho vs. mu-rho were plotted to distinguish lithologies and fluid distributions within the reservoir. Seismic interpretation identified three horizons (D-sand, C-sand, and B-sand) with a southeast-deepening trend and a shallower profile toward the northwest. The self-organizing map identified four main facies: sandstone, shaly sandstone, sandy shale, and shale. The N/M cross-plot analysis confirmed these classifications and revealed a mineral composition predominantly of quartz, validating the reliability and precision of the electrofacies results. Petrophysical interpretation identified two reservoir intervals in the Sawan-08 well and one in each of the Sawan-01 and Sawan-07 wells, all within the B and C sand levels. Rock physics modeling validated these findings by correlating predicted and actual wireline log data, confirming the precision of the model in estimating P- and S-wave velocities. Additionally, elastic parameter cross-plots effectively differentiated fluid types within the reservoir zones as wet sand, gas sand, shale, and shaly sand. This comprehensive methodology provides significant insights into the reservoir characteristics of LGF. It facilitates more precise hydrocarbon resource assessment, optimizes exploration initiatives, and refines reservoir management tactics, ultimately improving production in the area. The graphical abstract presents a structured workflow for reservoir characterization in the Lower Goru Formation (LGF) of the Sawan Gas Field, Pakistan, integrating seismic data (SEG-Y) and multi-scale well logs (caliper, gamma ray, sonic, resistivity, photoelectric, den-sity, and neutron). The study begins with seismic interpretation, where horizons were mapped using formation tops and synthetic seismograms, revealing a distinct southeast-deepening, northwest-shallowing trend in the D-Sand, C-Sand, and B-Sand units through structural contour maps. To ground these seismic observations in reservoir properties, petrophysical analysis was conducted, identifying two hydrocarbon-bearing intervals in the Sawan-08 well and one interval each in Sawan-01 and Sawan-07, primarily within the B-Sand and C-Sand levels. Cross-plots of log responses further constrained the lithology, confirming the reser-voir’s sandstone-dominated composition. Building on this lithological framework, an unsu-pervised machine learning approach using Self-Organizing Maps (SOM) was employed to classify electrofacies, delineating four distinct rock types: sandstone, shaly sandstone, sandy shale, and shale. To bridge facies classification with fluid dynamics, rock physics modeling was performed using elastic cross-plots (P-impedance vs. Vp/Vs and lambda-rho vs. mu-rho), which successfully discriminated lithologies and fluid types, including gas sand, wet sand, shaly sand, and shale. Together, this workflow unifies seismic stratigraphy, petrophysical evaluation, machine learning, and rock physics to better understand reservoir potential in the LGF.

This study examines factors that predict engagement with LinkedIn posts, specifically analyzing the impact of hashtags, tags, post age, and follower count on three engagement metrics: reactions, comments, and reposts. A negative binomial regression analysis of a random sample of 991 LinkedIn posts reveals that tags and hashtags significantly increase the expected number of reactions, with tags also substantially increasing comments. Follower counts slightly increase engagement, while post age negatively impacts expected counts across all metrics. The three engagement metrics are interrelated: comments boost reactions and reposts, reactions drive comments and reposts, and reposts increase reactions. These findings enhance our understanding of LinkedIn engagement and social media behavior by showing how certain message elements yield differing outcomes. Our findings also offer actionable insights for professionals and educators seeking to optimize their online presence and career outcomes on the platform.

Metacomputing optimizes distributed computing resources to enhance federated learning systems by enabling efficient resource allocation, improved scheduling, and greater scalability, thereby addressing challenges in large-scale and dynamic environments. This paper proposes an innovative framework integrating Directed Acyclic Graph (DAG) technology with federated learning within a metacomputing environment. The key contributions include a three-layer decentralized federated learning model integrating DAG and metacomputing to enhance resilience and scalability, two advanced tip selection models LazyEval Tip Selector and Precision Tip Selector to optimize node selection and improve data flow, and a Benchmark Improvement Protocol (BIP) for efficient node publishing and role adaptation.The BIP ensures that only high-performing models are published by comparing new models against established benchmarks, which enhances node collaboration and optimizes resource allocation. LazyEval Tip Selector minimizes redundant computations by leveraging a global cache and employing a lazy evaluation strategy, thereby improving computational efficiency. On the other hand, Precision Tip Selector uses a precise scoring mechanism to ensure accurate tip selection, thereby enhancing the robustness and reliability of the entire system. Collectively, these innovations enhance model training efficiency, support real-time updates, and improve the scalability of federated learning systems, making them well-suited for managing complex, dynamic environments.

Investigating the efficacy of goal-setting strategies is critical in understanding how individuals regulate their behavior, particularly within cognitive tasks. The present study examines the impact of self-set versus experimenter-set goals and point incentives on performance across three experiments using two sustained attention tasks. In Experiment 1, we compared self-set and experimenter-set goals in the psychomotor vigilance task, hypothesizing that self-set goals would lead to better performance due to increased agency. No significant differences emerged in task performance between the two conditions. Participants who self-set their goals also set increasingly easier goal standards over time. Experiment 2 introduced a novel task paradigm “Green Means Go,” modeled after the psychomotor vigilance task, and revealed faster reaction times in goal-setting conditions compared to a no-goal condition. Having a specific goal, either self-set or experimenter-set, was better for performance than having no goals. Experiment 3 allowed all participants to set their own goals and explored the influence of a points-based incentivization system on goal-setting tendencies. Those who received points set more difficult goals. Findings suggest that goal-setting mechanisms can enhance task performance and help reduce vigilance decrements, with potential implications for using goal-setting to elevate cognitive performance.

This letter demonstrates that scalar gauge functions give the most fundamental description of classical dynamical systems as such functions allow to construct Lagrangians for the systems, and to predict the future dynamical states of these systems, including their transitions from the periodic to non-periodic (chaotic) states, without solving the equations of motion.

Incomplete surgical resection in head and neck cancer can lead to locoregional recurrence in >35% of patients. Approaches such as image‐guided surgery (IGS) and post‐operative photodynamic therapy (PDT) have been proposed to reduce recurrence rates. However, the PDT doses needed to eliminate all unresected diseases are not established. This in vitro proof‐of‐concept study aims to predict head and neck tumor nodule viability in vitro following PDT with TLD1433 using the IGS probe ABY‐029. ABY‐029 is an EGFR‐specific affibody‐IRDye800CW conjugate that has undergone Phase 0 evaluation studies in head and neck cancer, among others. TLD1433 is a ruthenium‐based photosensitizer in a Phase II trial for non‐muscle invasive bladder cancer. Here, we demonstrate that decreases in fluorescence emission of ABY‐029 bound to MOC1 mouse head and neck cancer nodules in vitro can be predictive of TLD1433 PDT responses. Results show that photoactivation of TLD1433 produces reactive oxygen species (ROS) that reduce MOC1 nodule fractional viability in a manner that is inversely correlated with ABY‐029 fluorescence intensity (Pearson's r = −0.9148, R ² = 0.8369, p < 0.0001). We hypothesize that this is due to ROS‐mediated degradation of IRDye800CW. The findings warrant further studies using head and neck cancer nodules with heterogenous PDT responses and EGFR expression levels. If successful, the future goal would be to use ABY‐029 to guide the dosimetry of intraoperative PDT of the surgical bed after IGS to eliminate all microscopic diseases, reduce recurrence rates, and prolong survival.

Although ethical sensitivity is a well-defined concept, few scales specifically assess the ethical sensitivity of students in higher education referring to academic honor codes. We conducted the current study to develop and validate a scale that measures the Ethical Sensitivity Scale in Academic Activities (ESS). Scale development followed a standard survey design process with content validity. The final scale includes 20 items from three content areas: cheating, plagiarism, and falsification. Later, we validated the scale through factor analysis and measurement invariance at a national university of the U.S. with data from 654 students through a convenience sampling technique. After a thorough evaluation of alternative models, the one-factor structure was selected as the optimal model. Measurement invariance of gender and race was established. Furthermore, Rasch analysis provides evidence of scale (e.g., unidimensionality and monotonicity) and item quality (e.g., item fit; DIF). The results are relevant to stakeholders interested in examining and improving ethical sensitivity in academic activities for students in higher education.

The proper functionality of the brain depends on the ability of neurons to receive and transmit signals. Glial cells aid in signal transmission process by keeping the cells in healthy states. Additionally, in the event of any trauma, the glial cells impart support to the brain cells by isolating injured cells and keeping brain’s functionality active. Among the glial cells, the oligodendrocytes establish a direct connection with the neuron by encapsulating it with a fat-driven layer, termed a myelin sheath. This bubble of fat and protein provides insulation to the axons and assists in delivering electrical responses at an accelerated speed. Myelin sheath degeneration may occur due to sudden impact at the cellular level during brain trauma, which, may affect brain functionality due to impaired signal transmission along the axons. Here, using molecular dynamics simulations, we study the role of high strain rate mechanical loadings (108 and 109 s−1) on the damage threshold of myelin sheath. The molecular model was simulated using LAMMPS, an openly available molecular solver. The potentials for the associated system components were defined using OPLS force field. We have observed that upon mechanical loading, a multilayer myelin model can withstand up to 5–18% apparent tensile strain to failure, depending on loading sites and conditions. The separation of the protein from the lipid layers is considered to be the most likely failure mode of the myelin system.