Acute myeloid leukemia (AML) is caused by uncontrolled proliferation and impaired differentiation of hematopoietic stem and progenitor cells. Historically, research has emphasized the role of protein-coding genes in the development of AML. However, with the human genome project revealing that 98% of the transcriptome consists of non-protein-coding RNAs, recent studies have explored how the large classes of noncoding RNAs (ncRNAs) contribute to AML. Although there are many types of ncRNAs, much attention has been placed on understanding the function of long ncRNAs (lncRNAs) and small ncRNAs known as microRNAs (miRNAs). lncRNAs are >200 nucleotides, whereas mature miRNAs are typically 18 to 25 nucleotides. lncRNAs are involved in miRNA and protein sequestration and act as transcriptional and translational regulators, whereas miRNAs facilitate mRNA degradation and translational inhibition. In addition to lncRNAs and miRNAs, two additional types of ncRNAs, namely small nucleolar RNAs (snoRNAs) and circular RNAs (circRNAs), have recently garnered attention for their roles in AML. Here, we discuss how these four distinct classes of ncRNAs may aid in disease diagnosis and prognosis as well as the mechanisms by which their dysregulation contributes to AML.

It has been known for over a century that certain large classes of functions defined on a compact nowhere dense subset X X of the complex plane, and obtained as limits of analytic functions in various metrics, can sometimes inherit the property of unique continuation characteristic of the approximating family. The first example of the transfer of the uniqueness property in this way to R ( X ) R(X) , the space of functions that can be uniformly approximated on X X by a sequence of rational functions whose poles lie outside of X X , was obtained by M. V. Keldysh around 1940, but apparently never published. Years later in 1975 A. A. Gonchar exhibited a qualitatively definitive improvement of Keldysh’s example, and our goal here is to extend that result to R p ( X , d A ) R^p(X,dA) , p ≥ 2 p\geq 2 , the evidently larger space obtained as the closure of the rational functions in L p ( X , d A ) L^p(X,dA) , where d A dA denotes 2 2 -dimensional Lebesgue, or area, measure.

The shift towards HyFlex learning in the post-pandemic era has introduced new challenges for higher education, particularly in maintaining student engagement and motivation in online learning environments. This paper examines the potential of anonymous asynchronous online discussion forums (AODFs) to enhance participation and engagement in large online classes. We propose a new model of forum management that integrates question-answering and peer-to-peer interaction, allowing students to post questions anonymously while responses remain non-anonymous. The study investigates the evolving roles of teachers and students in promoting and participating in forum activities, adopting a “students as partners” perspective. Data from the implemented AODF indicate increased student participation and motivation, with a substantial portion of non-academic questions addressed through peer discussion. Challenges such as lurking behavior and the limitations of relying solely on technology are also highlighted. The study underscores the critical role of instructors in evaluating and adapting emerging technologies to meet student needs and foster a sense of community in online learning environments.

Black box warnings represent the most severe US Food and Drug Administration safety alerts. Currently, they are fragmented across individual product labels and not systematically accessible at the level of drugs or drug classes. This study provides a comprehensive overview of drugs with black box warnings and identifies class- and category-specific patterns of adverse effects, aiming to support risk benefit assessments and rational treatment. Six hundred twenty-six drugs across 250 drug classes were identified as carrying black box warnings. Drug classes were unevenly distributed, with many classes represented by a single drug, whereas a small number of large classes accounted for a substantial proportion of entries, for example, μ-opioid receptor agonists, cyclooxygenase inhibitors, angiotensin-converting enzyme inhibitors and kinase inhibitors. Across all drugs, 1016 adverse effect category listings were identified, demonstrating that multiple serious risks are frequently combined within a single boxed warning. Cardiac and cardiovascular adverse effects are the most prevalent category (18.7%), followed by immunologic and allergic reactions (11.4%), beside drug category-specific patterns. The standardized adverse effect count ranged from 1.1 in endocrine system drugs to 2.1 in cytotoxic treatments, indicating marked differences in the density and variety of serious adverse effects between drug categories. Black box warnings reveal pronounced heterogeneity across drugs, but show consistent, mechanism related patterns at the level of drug classes and categories. These patterns allow estimation of probable, high-risk adverse effects and risk profiles. A structured, drug class-based overview of black box warnings can improve awareness of safety risks and support rational prescribing. SIGNIFICANCE STATEMENT: Black box warnings are the most severe US Food and Drug Administration safety alerts, but they are fragmented across individual product labels. By systematically analyzing US Food and Drug Administration boxed warnings across drugs, drug classes, and adverse effect categories, this study reveals pronounced heterogeneity across individual drugs, but consistent, mechanism related risk patterns, with cardiovascular and immunologic adverse effects being most prevalent. A structured, drug class-based overview enables identification of high-risk profiles, supporting risk benefit assessment and rational prescribing.

This study aims to design and evaluate a Multimodal Generative AI Assistant capable of delivering real-time pedagogical feedback in large-scale computer science classrooms. The objective is to investigate whether integrating multimodal AI combining natural language processing, code analysis, speech interaction, and learning analytics can improve student learning outcomes, debugging efficiency, and engagement compared to traditional instructional support systems. The AI-assisted learning framework development and evaluation were based on Design Science Research (DSR) methodology. The system combines multimodal information of programming environments, natural language queries, and student interaction logs to produce contextual feedback based on generative AI models. A semester-long deployment was used, with 1,200 students, 600 in each control and AI-assisted group, using an A/B experimental design. Such quantitative measures as normalized learning gain, Time-to-Resolution (TTR), Hint Efficacy Score (HES), system latency, and multimodal synchronization accuracy were considered. The Technology Acceptance Model (TAM) and Subject Matter Expert (SME) audits were used to carry out a qualitative evaluation. The normalized learning gain of the AI-assisted group was found to be 0.73 versus 0.40 in the control group and statistically significant (p < 0.001), with a Cohen's d effect size of 0.91. The efficiency of debugging increased significantly, where Time-to-Resolution dropped on average by 134.5 minutes to 11.2 minutes (91.6% decrease). The system had a multimodal synchronization accuracy of 97.2% and a response latency of 1.65 seconds, and had a hallucination rate of 0.85%. Perceived Usefulness (6.6/7) and Perceived Ease of Use (6.3/7) were highly accepted. The results indicate that multimedia generative AI assistants can profoundly improve learning outcomes, decrease the debugging duration, and offer learning support during real-time instruction in large computer science classes at scale.

Academic writing is central to higher education, yet marking large cohorts of academic writing essays remains time-consuming. Assessors must align each essay with rubrics and provide detailed feedback, often under significant time constraints. This study examines whether artificial intelligence (AI) can streamline assessment of academic writing essays. Thirty first-year academic writing essays were assessed independently by an AI system and two experienced markers according to the Common European Framework of Reference for Languages (CEFR). The qualitative analysis involved comparing the nature and quality of AI-generated feedback and human markers, while the quantitative analysis compared the CEFR levels assigned by AI with those from the human markers. Findings show strong agreement between AI and human markers on surface-level errors (e.g., grammar and mechanics) but low agreement on CEFR proficiency classification. AI therefore appears well-suited for initial screening and formative commentary in large classes, while final proficiency judgements should remain with humans. The study contributes to debates on integrating AI in higher-education assessment by showing how AI can complement—not replace—human judgement, improving efficiency without sacrificing feedback quality. This exploratory study offers valuable insights; however, future research is warranted for deeper understanding of how AI can be integrated into higher-education assessment.

Abstract Digenean trematodes are parasites with a complex life cycle that often infest shell-bearing mollusks and produce distinct traces on the host skeleton that are recognizable in the fossil record. Here, three bivalve species ( Transennella conradina , Abra segmentum , and Chamelea gallina ) from Pleistocene and Holocene deposits of Florida and Italy were used to evaluate the hypothesis that trematode infestation affects shell morphology. The morphological effects of infestation were evaluated using geometric morphometrics and the pallial sinus index (PSI = pallial sinus length/shell length). For all three host species: (1) large size classes possess higher trematode prevalence (i.e., proportion of specimens possessing trematode-induced pits within a population) and higher per-specimen frequency of trematode-induced scars when compared with smaller size classes, suggesting ontogenetic accumulation of parasites; and (2) infested and non-infested specimens significantly differ in shell landmark-based morphology. Geometric morphometric analyses indicate that in two out of three species ( Transennella conradina , Abra segmentum ): (1) PSI and thin-plate spline analyses suggest significant pallial sinus reduction in infested specimens relative to non-infested; and (2) overall morphospace range, estimated by sample-standardized principal component (PC) hypervolume, was inflated with the inclusion of infested specimens. Consistent with previous studies, results indicate that trematode-induced morphological changes may influence the burrowing capabilities of the studied bivalves, affecting their ecological functioning and fitness. Changes in morphospace induced by trematode parasites hamper species delineation and confound morphometric and disparity patterns in the fossil record of infestation-prone species. Excluding fossil specimens with trematode traces can mitigate those confounding effects. Conversely, comparative morphometric analyses of infested and non-infested host specimens may allow us to investigate host responses to parasites over evolutionary timescales.

This conceptual paper introduces the responsive scaffold model (RSM), a theoretically grounded pedagogical and assessment framework designed to enhance soft skills development in construction management education. Using theoretical synthesis methodology, RSM integrates scaffolding theory, cognitive apprenticeship, and experiential learning into a cohesive framework. The model's theoretical foundations are well-established in educational literature, with each component drawing upon substantial research evidence from related contexts. Although developed within Australian higher education, the paper illustrates RSM's broader applicability across disciplines and international contexts. The model's responsiveness ensures tailored support that adapts to learners' evolving abilities, systematically reducing cognitive overload whilst maintaining academic rigour. Comparisons with problem-based and simulation-based learning highlight RSM's distinctive approach to balancing structured guidance and learner autonomy. Implementation guidelines cover curriculum design, faculty development, and authentic assessment, demonstrating RSM's scalability in large classes and digital settings. By aligning with educational standards and industry expectations, RSM offers a sustainable, learner-centred approach that bridges theoretical knowledge and real-world practice, advancing construction management education to enhance graduate employability and professional readiness. Future research directions are outlined to guide empirical validation of this conceptual framework.

The use of the Artificial intelligence (AI)-based instructional approaches in higher education has become a rapidly growing phenomenon. AI is being applied in universities in the following ways: adaptive learning, automated feedback, chatbots, intelligent tutoring, predictive analytics, and generative tools that can assist with content creation, explanation, and assessment preparation. In this paper, the author will discuss the effect of these AI-based modes of instruction on the quality of learning of students in institutions of higher learning. The concept of learning quality, in this provision, is a multidimensional phenomenon, which involves conceptual knowledge, interaction, timely feedback, self-regulated learning, critical thinking, academic achievement, and preparation to address real-life problems. The paper is of the view that AI can make learning much better in case it is implemented in the framework of good pedagogy, teacher coaching and ethical institutional policy. Recent systematic reviews demonstrate that AI could be used to facilitate personalization, improve student engagement, and make the instruction more responsive, especially when working with large and diverse classes. Nevertheless, some key issues, such as the excessive use of AI, academic dishonesty, algorithmic bias, lack of equal access, risks to data privacy, and the risk of superficial learning, are also mentioned in the literature. The paper also formulates research objectives and hypotheses, summarizes the recent literature and explains the pedagogical and institutional conditions in which AI produces the most positive impact. The results indicate that AI is not necessarily more likely to enhance the results of higher education, but its usefulness is conditional on its designing, governance, and implementation into the assessment and teaching processes. This paper concludes that an unbiased, morally controlled and evaluation conscious AI integration can be applied in the most effective way to achieve better quality of learning without compromising the critical thinking, equity and academic integrity in higher education establishment.

This study assessed the competence of teachers in teaching English in private English medium and public Kiswahili medium primary schools in Dodoma.The aim was to examine whether teaching practices aligned with the Tanzanian curriculum adopted in 2015, which emphasises quality education.Communicative Language Teaching (CLT) served as the theoretical framework.A mixed-methods approach with a convergent parallel design was employed.Data were collected through classroom observations, interviews, and questionnaires administered to English teachers of Standards 3 and 6.Purposive sampling was used to select teachers from four schools, and the data were analysed thematically and descriptively.The findings revealed that many teachers lacked competence in teaching English due to limited in-service training, short teaching experience, low language proficiency, inadequate resources, large classes, and limited awareness of CLT principles.The study concludes that these challenges hinder effective English teaching and reduce pupils' opportunities to develop communicative competence.It recommends strengthening in-service training, enhancing teachers' proficiency, providing adequate resources, reducing class sizes, and raising awareness of CLT to improve the quality of English teaching in Tanzanian primary schools.

In large-enrollment courses, instructors often receive hundreds of student comments that are difficult to synthesize, and traditional Likert-scale evaluations rarely capture the nuances of students’ learning experiences. This study introduces an accessible, evidence-based approach that uses open-source artificial-intelligence text analysis to interpret qualitative feedback quickly and systematically. Using 678 comments from 270 students in lecture and active-learning sections of an undergraduate course, the analysis identified clear contrasts in how students described engagement, learning, and enthusiasm - differences that were not visible in the numerical ratings. The AI results extended the insights from standard evaluations, highlighting aspects of classroom climate and instructional design that instructors could address directly. Because the procedure runs on common computers with freely available software, it enables faculty to examine large sets of comments in minutes rather than hours. This paper offers a practical guide for applying this approach to course evaluations, helping instructors turn extensive qualitative feedback into concise summaries for teaching improvement.

The objective of the study reported on here was to compare teaching performance in formative assessment between monograde and multigrade classes by class size, and to verify whether an association between teaching performance and class size existed in rural schools. A descriptive study was conducted on a representative sample of 539 elementary school students from rural schools in Chile. In monograde classes, better teaching performance was observed in students in large classes (83.3±12.3 points) compared to those in the medium (77.7±14.4 points) and small (77.4±14.4 points) classes (p < 0.01). In multigrade classes, the best teaching performance was observed in large classes (75.16±16.0 points), followed by medium (60.9±8.7 points) and small (61.6±18.4 points) classes. However, the results of a comparison between the 2 course modalities (monograde and multigrade) indicate better teaching performance in monograde classes than in multigrade classes. Furthermore, better teaching performance was observed in large classes in both modalities. However, the best teaching performance was observed in the monograde modality. In addition, we verified an association between teaching performance and class size.

Learner-centered feedback, emphasising future improvement, sensemaking, and student agency, has been increasingly recognised as effective, yet remains challenging for educators to offer in large classes with diverse learner needs. Recent advances in generative artificial intelligence (GenAI) offer new ways to support feedback practice, but limited research has examined how teachers use and perceive GenAI in real-world feedback practices, particularly for learner-centered feedback. Twenty-one higher education teachers were recruited to provide written feedback on a student presentation and then use a GenAI-powered feedback tool to analyse their feedback to identify learner-centered components (using BERT) and to generate enhanced drafts (using ChatGPT), and to explore their perceptions of its use. We analysed how teachers engaged with the BERT’s classification of teacher-written feedback text into learner-centered components, its suggestions to address missing components (which teachers could adopt or reject), and how teachers revised ChatGPT-enhanced feedback text. Findings indicated that teachers frequently adopted BERT’s suggestions and extensively revised ChatGPT outputs, often moderating praise, encouragement, and relationship-building statements. Interviews indicated that teachers valued GenAI for identifying missing components, improving language and structure, and promoting reflection, while also noting concerns about tone, trust, and the need for human editing.

The study is based on an Interpretative Phenomenological Analysis of the experiences of pre-service teachers undertaking a mandatory teaching practicum in Pakistan. Five pre-service teachers were enrolled in in-depth semi-structured interviews to give a portrait of their daily lives and struggles, and how they dealt with the struggles. Five themes of group experiences were found in the transcripts, which were theory-practice disjuncture, managing large classes, resource constraints, manoeuvring mentoring relationships, and managing workload stress. The main reason why the participants were deeply distressed by the thought of having to apply the pedagogical philosophies of the university in the classrooms where access to resources is minimal, there is constant overcrowding, and mentoring is unreliable. These findings show how teacher education in Pakistan shapes how pre-service teachers understand hardship as they construct professional identities in constrained conditions. This study can be valuable to the practicum research regarding identifying the challenges in resource-constrained scenarios, and the research implications of enhancing teacher education curriculum, strengthening the mentoring system, and enhancing the institutional support system in Pakistan, as well as in other emerging educational settings.

This study explores primary school teachers’ professional experiences in managing large classes, focusing on their characteristics, teaching and assessment practices, and the challenges they face. A critical case study approach using Atlas.ti 24, with semistructured interviews of 12 teachers (six male and six female) from two primary schools in Tanzania, was employed. Teachers developed their nurturing characteristics over time when teaching large classes. Teachers rely primarily on teacher-centered methods, such as lectures, corporal punishment discipline, and note-taking. Assessment practices include group tasks and summative tests; however, individualized assessments remain a challenge. Teachers face significant challenges, including insufficient resources, poor classroom environments, and limited government support. The small sample size limits the study's generalizability.

This study examined the relationship between class size and academic performance in Business Studies among junior secondary school students in Kosofe Local Government Area, Lagos State. Using a correlational research design, data were collected from 100 students across four public schools through archival records and perception surveys. Descriptive statistics revealed that class sizes were generally large, with some exceeding 90 students, while academic performance was predominantly average. Pearson correlation analysis indicated a significant negative relationship between class size and academic performance (r = –0.59, p < 0.05), suggesting that overcrowded classrooms adversely affect learning outcomes. Additionally, a positive correlation was observed between class size and perception scores, reflecting students’ awareness of the challenges posed by large classes. The findings underscore the need for policy interventions to reduce class sizes and improve instructional quality in skill-based subjects

Support Vector Machine (SVM) is a popular kernel-based method for data classification that has demonstrated high efficiency across a wide range of practical applications. However, SVM suffers from several limitations, including the potential failure of the optimization process, especially in high-dimensional spaces; the inherently high computational cost; the lack of a systematic approach to multi-class classification; difficulties in handling imbalanced classes; and the prohibitive cost of real-time or dynamic classification. This paper proposes an alternative method, referred to as Kernel-based Optimal Subspaces (KOS), which belongs to the family of kernel subspace methods. Mathematically similar to Kernel PCA (KPCA), KOS achieves performance comparable to SVM while addressing the aforementioned weaknesses. The method is based on computing the minimum distance to optimal feature subspaces of the mapped data. Because no optimization process is required, KOS is robust, fast, and easy to implement. The optimal subspaces are constructed independently, enabling high parallelizability and making the approach well-suited for dynamic classification and real-time applications. Furthermore, the issue of imbalanced classes is naturally handled by subdividing large classes into smaller sub-classes, thereby creating appropriately sized sub-subspaces within the feature space.

This qualitative research investigates how Communicative Language Teaching (CLT) functions in large tertiary-level ESL classrooms from the viewpoint of teachers. The study was carried out at a university in Pakistan, where overcrowded, examination-driven classes are a common reality. Information was gathered through semi-structured interviews with six English language instructors and examined using thematic analysis. The results indicate that most teachers regard CLT as an effective, student-focused approach that promotes learners’ speaking ability, fluency, and self-confidence by involving them in meaningful, real-world communication activities. Despite these advantages, participants also identified several practical difficulties in implementing CLT in large classes. These challenges include managing and supervising all students, providing personalized feedback, handling classroom noise and discipline, coping with limited time, and addressing unequal participation between high- and low-performing students. To overcome these issues, teachers reported modifying CLT practices by incorporating pair and group work, assigning specific roles to students, designing shorter and more structured activities, and selecting locally relevant topics. They also combine CLT with other teaching approaches, such as the Grammar-Translation Method, the Direct Method, Task-Based Learning, and Audio-Lingual techniques. The study emphasizes the importance of professional development, institutional backing, improved class management or reduced class sizes, and assessment systems that prioritize communicative competence. Overall, the findings suggest that while CLT can positively influence large ESL classrooms, its effectiveness depends on flexible, blended implementation that takes contextual limitations into account.

Blended cooperative learning combines online digital tools with face-to-face group activities to foster interactive, student-centered education, integrating the flexibility of blended learning with the social and cognitive expansions of cooperative structures to enhance engagement, critical thinking, and peer collaboration. Despite its promise, implementation comes across significant challenges. Technological obstacles, including unreliable internet, limited device availability, and the digital divide, hinder equitable access and participation, especially in underserved areas. Teachers face increased workloads from redesigning curricula, facilitating hybrid group dynamics, etc., which is often intensified by inadequate training and low self-efficacy in managing blended environments. Cooperative processes are interrupted by unequal contributions, interpersonal conflicts, insufficient student social skills, and challenges maintaining accountability. Classroom management grows difficult amidst noise, time pressures, large classes, and tensions between individual and collective assessment. Students struggle with poor self-regulation, weakening motivation during asynchronous phases, isolation, and distractions from non-academic online content. Institutional barriers such as rigid curricula, resistance to change, and weak policy support make adoption even more difficult. Targeted teacher training, fair resource distribution, specific education in collaborative skills, organized group responsibilities, and definite evaluation procedures are all necessary to meet these objectives. When blended cooperative learning is properly implemented, it may produce lively and inclusive classrooms that prepare students for better teamwork in real-world situations.

We construct explicit strictly ascending chains of dense subalgebras of length 𝔠 in every separable infinite-dimensional complex Banach algebra. For large classes of commutative C*-algebras we also construct strictly descending chains of the same length. The constructions rely on algebraic independence, Stone–Weierstrass arguments, and transfinite recursion.