Interpretable machine learning and greenness assessment for sustainable tetracycline adsorption and phytotoxicity mitigation using almond peel - derived activated carbon.

Synthetic data-augmented and machine learning assessment of creep life in adhesively bonded joints

Validity evidence for the communication learning assessment: Prior coursework as a known group.

Saliency-guided video coding via recurrent learning and perceptual quality assessment

The extensive consumption of cement-based materials and the growing accumulation of plastic waste present serious environmental challenges due to high carbon emissions and long-term pollution. To address these issues, this study proposes a cement-free construction approach in which cement is completely eliminated, and waste plastic is utilized as a binding material to develop plastic–sand composites. Unlike existing studies that primarily focus on partial cement replacement, this research integrates composition-based material development, basalt fibre reinforcement, flexural-strength (F-S) characterization, and explainable machine learning within a unified framework. A comprehensive experimental programme was conducted to evaluate the influence of sand size, plastic type and content, and basalt fibre reinforcement on F-S. Experimental results showed that finer sand produced higher F-S (up to 4.26 MPa), commercial plastic-based composites outperformed waste plastic composites (maximum 4.91 MPa), and the incorporation of basalt fibres significantly enhanced F-S by approximately 35 %. Data-driven models, including support vector regression (SVR), extreme gradient boosting (XGB), and gene expression programming (GEP), were developed using the experimental dataset. Model evaluation using statistical indicators and k-fold cross-validation demonstrated high predictive accuracy, with XGB achieving the best performance (R² = 0.99, RMSE = 0.113 MPa). Explainable artificial intelligence analysis using SHAP quantified the relative contribution of each input parameter, identifying basalt fibre content and sand size as the most influential parameters governing flexural behaviour, while GEP provided empirical equations for transparent strength prediction. The proposed framework demonstrates the feasibility of cement-free composites combined with explainable AI for sustainable and performance-oriented construction materials.

Neurocognitive functions play a fundamental role in children's learning and academic success. Among these functions, memorization-based on associative cognitive processes-is essential for knowledge acquisition. In Morocco, educational disparities between rural and urban environments and differences between public and private schools may influence cognitive development and learning outcomes among schoolchildren. This study aimed to analyze the memorization process among Moroccan school-aged children and to identify demographic and contextual factors associated with learning performance. A cross-sectional study was conducted among 441 children aged 8-15 years recruited from six public and private schools in the Marrakech-Safi region of Morocco between November 2022 and March 2023. Memorization and learning abilities were assessed using the adapted Moroccan version of the Rey-Taylor auditory-verbal learning test (AVLT). Key indicators included the total number of recalled words across 10 trials, the learning index, and retroactive interference. Associations between cognitive performance and contextual variables (age, gender, school type, and residence) were analyzed using Spearman correlation tests. The results show that the learning index is relatively higher among girls (M=4.8, standard deviation [SD]=2.70), children over 12 years old (M=5.0, SD=2.39), children from rural areas (M=5.0, SD=2.62), and those enrolled in the public sector (M=5.0, SD=2.59), compared with other categories. These variables were the least affected by retroactive interference except for the rural environment, which experienced an RI=68.8, SD=0.21. However, the data indicate a performance among private-sector and younger students in at least one learning assessment stage. The Spearman test showed a positive correlation between the learning index and the age variable (r=0.095; p=0.046), and between retroactive interference and the living environment (r=0.105; p=0.028) and the type of school (r=0.126; p=0.0008). Overall, improving memory and academic performance requires targeted pedagogy, extracurricular activities, equal school opportunities, and research on factors, such as motivation, social dynamics, and teaching methods. Similarly, longitudinal studies will help deepen the understanding of memory and guide educational policies.

Traditional Bayesian network structure learning requires explicit trade-offs among structural complexity, statistical fit, and decision reliability. This study proposes BAPS (Budgeted Acyclicity with Phase-transition Spectral diagnostics), a framework that reformulates structure learning as a budget-controlled optimization problem in which a global edge budget limits the number of selected candidate edges and thereby regulates model capacity. A relaxed QUBO formulation with post-hoc DAG projection explores likelihood-improving structures while recovering feasible directed acyclic graphs; in benchmark experiments, the repair process reduces SHD by 20.6%, 14.9%, and 8.3% on Asia, Insurance, and Barley, indicating improved structural agreement while enforcing feasibility. Spectral diagnostics based on algebraic connectivity serve as descriptive structural indicators for flagging budget regions where repair burden may escalate. A dual-layer credibility framework quantifies uncertainty from parameter and observational sources; credibility interval width contracts by 69-73% under increasing Dirichlet concentration and remains empirically stable in sparse-data conditions where resampling-based methods become unstable due to zero-frequency effects. Across benchmark networks, BAPS achieves the highest BIC gain while maintaining broadly comparable held-out predictive performance relative to established baselines. Overall, BAPS provides a unified framework integrating edge-budget capacity control, structural diagnostics, feasibility repair, and credibility assessment for Bayesian network learning in complex diagnostic environments.

ABSTRACT This exploratory qualitative study investigates how creative mathematical story writing can illuminate primary students’ geometric conceptions and misconceptions in an Irish‑medium context. Motivated by persistent underperformance in geometry on international assessments and an over‑reliance on often problematic textbooks, the study examines whether creative writing can make students’ concept images of geometry visible. Drawing on the Fighting Words pedagogy, 18 students (ages 10–12) collaboratively authored four mathematical stories in Irish, later translated into English. Using qualitative content analysis, we coded story excerpts against Irish curriculum objectives for Geometry and Geometric Measurement, with conceptual understanding serving as the primary analytic lens. Findings indicate that students displayed substantial procedural fluency (e.g. correct formulas for perimeter, area, and circle measures; accurate classifications of shapes) while simultaneously revealing fragile concept images. Recurring misconceptions included conflating 2‑D figures and 3‑D solids, interpreting 360° as a “sharp turn” rather than a full rotation, and treating area, volume, and capacity as interchangeable measures of “size.” The study argues that creative mathematical story writing offers a powerful diagnostic window on students’ geometric thinking, particularly in minority‑language settings, and discusses implications for assessment for learning and for designing tasks that productively surface and address misconceptions.

ABSTRACT In Iran, a shift in educational policy was introduced 15 years ago, placing a stronger emphasis on assessment for learning (AfL). However, this shift has not been fully embraced by teachers of English as a Foreign Language (EFL) because of various factors including limited training in assessment and exam-oriented practices. This study explores how a professional development program with a strong emphasis on integrating theory and practice and teacher co-construction of knowledge can support teacher development in assessing literacy and implementing AfL in the classroom. The one-month professional development program included four 90-minute online workshops and teacher interactions in the forum, where they responded to the intervention assignments and reacted to their colleagues’ responses. We traced changes in teachers’ understanding of classroom assessment and how they could merge AfL and assessment of learning (AoL) in their classroom. Data included pre- and post-intervention open-ended questionnaires and teachers’ forum interactions. We present the unique developmental trajectories of two EFL teacher participants to build an argument for praxis, a dialectical partnership between researchers/teacher educators and teachers, which allows these trajectories to emerge. The implications of the findings of this study are discussed with reference to teacher training programs and in-service teacher training.

Educational transformation in the era of artificial intelligence requires schools to build a humanistic-digital, reflective, collaborative, and data-driven learning ecosystem. This study aims to map the implementation quality of deep learning in State Senior High Schools adopting the Deep Learning, Coding & Artificial Intelligence (DLCAAI) model in Central Java and formulate its implications for evidence-based policy development (baseline assessment), involving 316 teachers from all DLCAAI model schools in the region. The research instrument was a 5-point Likert-scale questionnaire with 75 items covering deep learning principles, learning experiences,t he deep learning framework, graduate profiles, and learning assessment. Validity tests showed all items were valid (corrected item-total correlation > 0.30), while instrument reliability was very high (Cronbach’s Alpha = 0.992). Data analysis employed means, percentages, categories, and comparative dimensional analysis. Results indicate that the quality of deep learning implementation is 93.43%. The strongest dimensions include student-centered learning, psychological safety, character development, inclusivity, and digital integration. However, metacognitive reflection, formative assessment, learning differentiation, cross-disciplinary learning, and the utilization of learning data still require strengthening. The novelty of this research lies in developing the Baseline Architecture of Humanistic-Digital Deep Learning Ecosystem framework, which integrates deep learning, AI, reflective pedagogy, student agency, and evidence-based educational policy. This study confirms that baseline assessment functions as an educational diagnostic, a compass for school transformation, and a policy navigation system in the humanistic and sustainable AI-based educational transformation.

Abstract Landslides are significant natural hazard in the Tehri Garhwal District in Uttarakhand, India, posing risks to human life, infrastructure, and environment. As the district ranks second among 147 districts based on the landslide index, accurate spatial modelling of landslide susceptibility is crucial for effective risk management. This study presents a comprehensive comparative assessment of machine learning (ML) and deep learning (DL) models for spatial landslide modelling. The study further examines the varying correlations between conditioning factors and the occurrence of landslides. The models were trained and tested using a 70:30 split, with 17 landslide conditioning factors and 1,600 landslide locations. Model performance was assessed using the Area Under the Receiver Operating Characteristic Curve, mean absolute error, and root mean square error. The novelty of this study lies in the integrated evaluation of conventional machine-learning and deep-learning models under a unified framework using the most recent and updated landslide inventory and conditioning datasets for the study area. The accuracy of the deep learning neural network models was the highest (90.31%), using the most recent datasets. The outcomes of this study provide valuable insights for researchers and practitioners in the development of reliable landslide susceptibility maps for evidence-based planning and risk mitigation in mountainous regions.

Efficacy and hepatotoxicity of Psoralea corylifolia L.: A meta-analysis and machine learning assessment in postmenopausal osteoporosis.

ABSTRACT Effective instructional decision-making in mathematics requires timely and accurate assessment of student learning. This study provides validity and reliability evidence for an adaptive mathematics assessment tool designed to support teachers in identifying areas where students need instructional support. A total of 4,875 Chilean students from fifth to eighth grade participated in the calibration of an item bank. Validity evidence was collected from four sources: test content, response processes, internal structure, and relationships with other variables. Reliability was assessed using a test-retest approach. The results provide validity evidence and reliability estimates that support the use of the tool for formative purposes. The assessment is well-suited to the Chilean educational context, offering a practical solution for evaluating mathematical learning and informing pedagogical interventions. Furthermore, the tool’s framework serves as a model for developing adaptive mathematics assessments that can be adapted for use in other Spanish-speaking regions or countries seeking data-driven instructional practices.

The goal of the current study is to close a significant gap in the literature about the representation of concrete produced from industrial waste. In example, prior research experiences low levels of model explainability, limited generalizability, and limited data, especially because of their inability to capture nonlinear interactions among features. In order to overcome these limitations, authors apply seven supervised machine learning models to a large dataset of 711 observations. The two main new features of the suggested strategy include a thorough, systematic assessment of ensemble learners and SHAP-based model interpretability. With a maximum test accuracy of R2 = 0.881, RMSE = 5.65MPa, MAE = 4.17MPa, and MAPE = 25.15%, the Gradient Boosting Regressor (GBR) outperformed other cutting-edge models such as CatBoost (R2 = 0.879) and Histogram Gradient Boosting (R2 = 0.878). Based on the findings from the SHAP analysis, it is clear that the machine learning model shows more sensitivity to the fine aggregate content than to cement and other materials, making it the most influential factor in the global model. Water content and curing age follow. In actual use, the suggested model and its intuitive Python-based graphical user interface (GUI) can quickly and affordably estimate compressive strength without requiring expensive and time-consuming laboratory testing, guaranteeing the successful incorporation of industrial waste products into the concrete mixture.

PurposeMonitoring activities of daily living (ADLs), such as walking, sitting, and stair climbing, is an important indicator of functional status, autonomy, and overall well-being in older adults. Traditional assessment approaches, such as questionnaires or offline machine learning models, struggle to adapt to dynamic environments and to the natural variability of human movement. This study aims to design, implement, and evaluate a real-time data collection and processing system that supports the training and assessment of Stream Learning (SL) models for ADL classification, and to compare their performance with conventional offline models. This study aims to design, implement, and evaluate a real-time data collection and processing system for ADL classification using Stream Learning (SL) models. Additionally, the study analyzes whether SL-based monitoring can provide more reliable and responsive metrics than traditional offline approaches under real-world conditions.Materials and MethodsThe study involved nine adult participants during offline and online experimental phases aimed at evaluating real-time ADL monitoring under controlled and uncontrolled conditions. A mobile application captured accelerometer and gyroscope signals from smartphones and transmitted the data streams to a cloud server for window segmentation, label alignment, storage, and incremental model updates. Both offline models and SL algorithms were evaluated using precision, adaptability, and prediction stability metrics. The data captured were obtained through experiments approved by an Ethical Committee, and those involved signed an informed consent.ResultsThe system successfully enabled continuous real-time data acquisition and incremental model training. Offline models achieved competitive accuracy but limited adaptability to variations in movement patterns or acquisition conditions. SL models demonstrated more robust and stable predictions, adapting better to the natural variability of daily activities in real-world scenarios.ConclusionThe proposed system demonstrated the feasibility of integrating mobile sensing and SL for continuous ADL monitoring under real-world conditions. Compared with traditional offline models, SL approaches achieved superior adaptability, stability, and responsiveness during continuous operation, with Hoeffding Tree showing the best overall balance between accuracy and latency. These findings support the potential of SL-based systems for scalable and low-cost functional monitoring applications. Further validation using larger and clinically characterized populations is still required.

Multimorbidity, the presence of multiple chronic conditions in one person, is a growing global health concern. Integration of chronic care services is urgently needed, especially in low-resource settings including in Southern Africa, where care has been fragmented by vertical and siloed disease approaches. Many countries share similar challenges of integration, presenting rich opportunities for shared learning. Yet, rarely are these opportunities capitalised upon, in part because of a lack of systematic knowledge about the similarities and differences in health system contexts, challenges and current progress towards integration. As part of an inter-country collaboration, we sought to answer the questions: What are the common and distinct characteristics of the care pathways for people living with multimorbidity in Malawi and Zimbabwe, and the opportunities and challenges that emerge through such a country-level comparison? We used an iterative, qualitative research design that involved a desk review of relevant indicators, policies and strategies; key informant interviews, collaborative workshops, and the development of case studies of service integration in practice. Thematic analysis and comparison of challenges of integration across different levels of care revealed uneven funding for different diseases, a lack of both ‘vertical’ and ‘horizontal’ integration, frequent stockouts of drugs and diagnostic equipment, especially for noncommunicable diseases (NCDs), and inadequate training and support for clinicians. In both countries, progress towards decentralising and integrating chronic disease care at national level, has occurred through inclusion of specific NCDs into HIV programmes. This is prone to leave out comprehensive chronic care for people that are not living with HIV and reproduces verticalised programming. We suggest that a promising avenue for wider scale-up of decentralised, non-HIV-dependent integrated care lies in the expansion of an Integrated Chronic Care Clinic (IC3) model that provides comprehensive health system integration for all chronic diseases. Further cross-country learning and feasibility assessment is needed to advance this model.

Conventional physical education assessment depends on subjective teacher observation and simple rating scales, often resulting in subjective bias, low evaluation efficiency, and delayed instructional feedback. This study constructed a hierarchical multimodal fusion framework integrating inertial measurement unit (IMU) data and expert rubric scores for standardized PE skill evaluation. The framework adopts Dynamic Time Warping for spatiotemporal alignment to resolve asynchrony between continuous sensor signals and discrete manual scoring, and applies adaptive gated fusion to balance modality weights according to data quality. A total of 3,920 skill samples from 245 middle-school students across three schools and four sports were independently annotated by three calibrated PE teachers, achieving high inter-rater reliability (ICC(2,k) = 0.87, 95% CI [0.84, 0.90]). All experiments were conducted with participant-wise data splitting and ten repeated random-seed trials to ensure result stability. The proposed multimodal model achieved an overall accuracy of 91.3 ± 0.4%, significantly surpassing single-modality baselines (p < 1 × 10⁻⁸). Spatiotemporal alignment effectively eliminated temporal mismatch, and the model maintained stable performance under simulated data missingness and sensor failure. The 43.2 ms single-sample inference speed meets real-time classroom assessment demands. Ablation experiments verified the essential role of cross-modal attention in multimodal learning. Leave-one-school-out and cross-stratum evaluations confirmed stable performance across different schools, genders, age groups and skill levels. The current multi-validation results support reliable within-population model performance, while external validation with independent cohorts is still required for broader generalizability. Model interpretability analysis further validated the biomechanical rationality of the learned assessment patterns. This study provides a feasible multimodal digital assessment pipeline for standardized and formative physical education evaluation. The system intelligently reproduces expert-level teacher scoring rather than objective biomechanical measurement, which makes it a powerful auxiliary tool for classroom teaching rather than a substitute for professional teacher judgment. This work offers a practical digital transformation pathway for PE assessment, with standardized curriculum alignment and staged teacher professional development supporting reliable real-world educational deployment.

To develop a structured, pharmacy-specific framework that defines the essential digital health and telehealth knowledge and skills required for student pharmacists and practicing pharmacists across educational and practice settings. An expert panel convened by the American Society of Health-System Pharmacists conducted a targeted literature review and environmental scan, followed by an iterative, consensus-driven development process over 18 months. The resulting framework organizes competencies into 7 domains: Technology and Digital Literacy; Patient Communication, Education, and Engagement; Documentation and Information Management; Clinical Assessment and Decision-Making; Operational Management and Workflow Integration; Regulatory Compliance and Ethical Practice; and Quality Improvement and Evaluation. Each domain includes actionable skills aligned with real-world telehealth workflows and mapped to entrustable professional activities to support competency-based education. The framework also incorporates practical curricular applications, including learning objectives, assessment strategies, and experiential activities, to facilitate integration into didactic curricula, experiential training, and continuing professional development. This approach addresses gaps in current pharmacy education, where digital health training is often fragmented, inconsistent, or insufficiently tailored to the pharmacist's role in interdisciplinary care. This framework provides a comprehensive and adaptable foundation for integrating digital health and telehealth competencies into pharmacy education and practice. By standardizing expectations and aligning training with evolving healthcare technologies, it supports the preparation of pharmacists to deliver safe, effective, and patient-centered care in digitally enabled environments while promoting ongoing professional development and workforce readiness.

This mixed-methods study compared ChatGPT-4 generated feedback with teacher feedback in assessing secondary school EFL learners’ writing. Conducted in a state secondary school in Istanbul, Türkiye, the study involved 53 fifth-grade students who completed weekly writing tasks over six weeks. One class received teacher feedback, while two classes received feedback from ChatGPT-4. Students revised their texts based on the feedback, and changes in writing performance were analysed quantitatively. In addition, semi-structured interviews with 45 students were conducted to explore their feedback preferences and perceptions. The quantitative findings showed that both feedback types supported improvement in students’ writing, although teacher feedback produced more consistent gains across the six-week period. A strong positive correlation was found between ChatGPT-4 and teacher scores in both pre- and post-feedback assessments, suggesting that AI-generated scoring aligned closely with human evaluation. However, the qualitative findings revealed that students generally preferred teacher feedback, describing it as more personal, motivating, and easier to trust. ChatGPT-4 feedback was appreciated for its speed, clarity, and accessibility, but was also seen as less detailed and less emotionally engaging. The findings suggest that generative AI can serve as a useful formative feedback tool in EFL writing instruction, but it does not replace the pedagogical and relational strengths of teacher feedback. A hybrid model that combines the efficiency of AI with the contextual sensitivity of human feedback may offer the most effective approach for supporting writing development in secondary language classrooms.

Hepatitis, a disease characterised by inflammation of the liver, is a leading global health challenge that contributes to over 1.3 million deaths annually, with hepatitis B and C accounting for many of these fatalities. Intensive care unit (ICU) management of patients is particularly challenging due to the complex clinical care and resource demands. Despite advancements in ICU predictive analytics, limited research has specifically addressed hepatitis patients, creating a gap in optimising care for this population. This study focuses on predicting ICU length of stay (LoS), hospital discharge outcomes and discharge location for ICU-admitted viral hepatitis patients using a comparative assessment of machine learning (ML) models. Leveraging data from the Medical Information Mart for Intensive Care-IV database, which includes around 94 500 ICU patient records, this study uses sociodemographic details, clinical characteristics and resource utilisation metrics to develop predictive models such as Random Forest, Logistic Regression, Gradient Boosting Machines and Generalised Additive Model with Negative Binomial Regression. Among 3875 ICU-admitted hepatitis patients, Random Forest classification outperformed Logistic Regression in predicting discharge outcomes, achieving higher accuracy (0.87 vs 0.82) and greater discriminative ability (area under the receiver operating characteristic curve 0.95 vs 0.89). For ICU LoS prediction, Random Forest regression applied to log-transformed LoS demonstrated strong performance (R² up to 0.82), while the generalised additive model with negative binomial distribution explained approximately 76% of LoS variance. Prediction of discharge location yielded moderate performance across Gradient Boosting and multinomial logistic regression models (accuracy 0.55 and 0.56), reflecting challenges associated with multi-class imbalance. Variable importance analyses across ML models consistently identified medication counts, procedure counts, comorbidity burden, age, race and total LoS as the most influential predictors of discharge outcomes and discharge location. This study demonstrates the value of ML models for predicting clinical outcomes for hepatitis patients, including ICU LOS and hospital discharge status. The results underscore the influence of factors like race and age, revealing disparities that must be addressed in predictive care strategies. While the models show promise, challenges such as variability in prolonged stays and limited multiclass prediction accuracy point to the need for ongoing refinement and research.