Rehabilitative therapies play a crucial role in upper limb motor recovery, as upper limbs are the most active parts in executing the activities of daily living. Because of a huge number of people with motor disorders and a shortage of therapists, the integration of data-driven Artificial Intelligence methodologies and robots for rehabilitation could be helpful in creating personalized and challenging therapies, leading to a myriad of benefits for both patients and therapists. Artificial Intelligence methods can be implemented in different functional modules of the robotic platform, such as user intention recognition, robot motion planning, robot interaction control, and system adaptation through different learning paradigms. This paper presents a systematic literature review on the use of data-driven learning methods applied in upper limb robot-aided rehabilitation. The analysis is structured around the learning paradigms adopted, namely, supervised, unsupervised, and reinforcement learning, as well as the corresponding task types (e.g., classification, regression, and control tasks) and model types, distinguishing between machine learning and deep learning approaches. The review reveals that most studies employ supervised learning to address classification tasks, and that deep learning models are the most frequently adopted.

Capturing peer discourse in real-world classrooms offers valuable insights into collaborative learning but presents significant technical and pedagogical challenges. While most existing Automatic Speech Recognition (ASR) systems research has focused on teacher-led or online English-speaking environments, peer-to-peer dialogue in noisy, non-English dominant, face-to-face classrooms remains underexplored. This study investigates the feasibility of using both traditional ASR systems—Whisper and Wav2Vec2—and emerging Large Audio Language Models (LALMs), including Qwen2-Audio and Ultravox, to transcribe Mandarin peer conversations recorded via students’ mobile phones in authentic classroom settings. We collected over 105,715 seconds of audio from 38 student groups across two collaborative learning tasks from university classrooms. The manually transcriptions were served as ground truth. Audio quality test of all audio recordings was conducted. Five representative samples with varied signal-to-noise ratios (SNR) and speech ratios were selected to do in-depth analysis. Transcription quality was evaluated using Word Error Rate (WER), Character Error Rate (CER), and Fuzzy String Matching. Additionally, we conducted a thematic analysis of transcription errors to identify linguistic, acoustic, and task-related challenges. Results show that Whisper consistently outperforms other models, achieving high transcription fidelity even in moderately noisy conditions. In contrast, LALMs—despite their strengths in semantic understanding—performed poorly in verbatim transcription, often generating hallucinated or irrelevant content. Importantly, task type and speech characteristics significantly influenced model performance: structured, reflective discussions yielded better results than spontaneous, technical dialogues involving numeric and English domain terms. This study contributes a low-cost, replicable workflow for classroom audio collection and evaluation, along with a detailed taxonomy of transcription errors. We emphasise that our results are exploratory due to the limited sample size. Nevertheless, the findings highlight the current limitations of LALMs for ASR tasks and offers practical recommendations for model selection in educational contexts. Our findings support the responsible integration of ASR technologies into classroom practice, with implications for real-time feedback, collaborative learning analytics, and teacher professional development. For researchers, this work demonstrates the need to consider peer dialogue and multilingual classroom ecologies when evaluating ASR. For teachers, practical recommendations are offered for selecting transcription tools that can support real-time feedback and professional reflection. For lifelong learning, our study illustrates the potential of ASR technologies to make collaborative dialogue more visible, analysable, and actionable across diverse contexts.

Humans prepare motor actions when perceiving objects that afford specific behaviors, highlighting the tight link between perception and action. For example, seeing a graspable object like a mug can trigger hand movements aligned to its handle - a phenomenon known as the object affordance effect. Vainio et al. (Quarterly Journal of Experimental Psychology 64, 1094-1110, 2011) demonstrated this can produce a negative compatibility effect (NCE). This occurs when a spatially compatible prime object eliciting an affordance (e.g., a mug), but to be ignored, precedes a target requiring a spatial response. Given that task demands shape response execution (e.g., Schöpper & Frings, Attention, Perception & Psychophysics, 86, 171-185, 2024), we hypothesized that the effect of affordance would vary accordingly. In Experiment 1, participants performed three tasks: arrow direction discrimination, shape discrimination, and circle localization. In all tasks, the time interval between the affordance object (a mug) and the onset of the target, as well as the compatibility between the mug and the response, varied. The arrow task replicated the NCE - responses were slower in compatible trials at short intervals. No compatibility effects were observed in the shape task. Notably, the localization task revealed a positive compatibility effect (PCE). The variation in compatibility effects suggests task-dependent affordances. Experiment 2 manipulated the target position relative to the fixation to investigate the PCE in the localization task and explore the differences in the compatibility effect. Although the PCE was not replicated, the NCE now also appeared for location tasks. Our results suggest that task constraints shape the compatibility effect, and distractor-induced affordances engage inhibitory mechanisms only when spatial features are relevant.

Tracking self-regulated learning in action: How individual differences shape positive and negative regulation across three types of tasks

In full-touch HMI mode: How does car-following pressure, task complexity, and speed affect driver's visual distraction characteristics?

Task-dependent neural effects of physical exercise: A systematic review and meta-analysis of functional magnetic resonance imaging studies.

Abstract. Problem. The modern educational space, which has long crossed national borders, faces significant challenges concerning quality, goals, content, and structure. Unless these challenges are addressed, it is difficult to ensure the professional competence of graduates from technical universities and other higher education institutions. Therefore, one of the key tasks of the modern higher education system is to align the quality of professional training with the evolving needs of individuals, society, and the labour market. Goal. This study aims to explore the development of high-quality professional training for graduates of Ukraine’s technical higher education institutions to meet the requirements of the modern labour market. The authors propose a graduate model for technical universities, based on personal qualities, the needs of industry, and the professional competences required for addressing real production tasks. Methodology. Through the analysis of the interaction between the higher professional education system and the labour market, the authors examined, on the one hand, the professional and personal competences required by employers, and on the other hand, the standards of higher education. The comparative analysis of different respondent groups regarding the quality of training in technical universities led to the conclusion that the model of a modern technical university graduate should include the model of the specialist’s personality, the model of their professional activity, and the model reflecting the demands of the modern labour market. The main requirements for such a graduate should be determined by their functional responsibilities, the tasks of a specific workplace, and the type of production tasks they encounter. Results. Developing high-quality professional training for graduates of Ukraine’s technical higher education institutions, in line with the requirements of the modern labour market, is an urgent task not only for universities but, first and foremost, for employers. They need to define specific requirements for the specialists they expect, rather than general ones. Only this approach to shaping the graduate model will enable technical universities to train engineers who do not require a long period of adaptation to real production conditions or retraining to acquire the competences demanded by a particular enterprise.

MMSupcon: An image fusion-based multi-modal supervised contrastive method for brain tumor diagnosis.

Figurative language production in schizophrenia, bipolar disorder, and depression.

Over the last decade, the animacy effect has emerged as a significant advantage in the processing of adaptive memory, illustrating that individuals tend to have superior memory for animate over inanimate stimuli. Despite this, a systematic analysis of the effect remains absent. A quantitative review is needed to assess the stability of the animacy effect, and the reverse animacy effect observed in some studies also requires investigation. Employing a three-level meta-analytic approach, this study provides a comprehensive evaluation and synthesis of the animacy effect's influence on memory processing. Through the integration of 45 primary studies, we conclusively demonstrate the consistent presence of the animacy effect within the context of enhanced memory processing, characterized by a large effect size (ηp2= .19, 95% CI [.38, .55]). Our findings indicate that the impact of the animacy effect is robust across variations in study year and geographical location, confirming its stability across diverse cultural and temporal frameworks. Importantly, our analysis revealed that the animacy effect was moderated by the type of memory task. Specifically, the animacy effect was stronger in free recall compared to recognition and cued recall, with the latter two showing a less consistent animacy effect. This insight underscores the necessity of considering the memory task type in research on the animacy effect, particularly in experimental designs investigating the underlying mechanisms of its influence. In sum, although the magnitude of the animacy effect may vary across memory tasks, it represents a stable memory advantage shaped by evolutionary pressures. Continued research is needed to uncover its cognitive underpinnings and to translate these findings into practical domains such as education and marketing.

Graph data in real-world scenarios undergo rapid and frequent changes, making it challenging for existing graph models to effectively handle the continuous influx of new data and accommodate data withdrawal requests. The approach to frequently retraining graph models is resource intensive and impractical. To address this pressing challenge, this paper introduces a new concept of graph memory learning. Its core idea is to enable a graph model to selectively remember new knowledge but forget old knowledge. Building on this approach, the paper presents a novel graph memory learning framework - Brain-inspired Graph Memory Learning (BGML), inspired by brain network dynamics and function-structure coupling strategies. BGML incorporates a multi-granular hierarchical progressive learning mechanism rooted in feature graph grain learning to mitigate potential conflict between memorization and forgetting in graph memory learning. This mechanism allows for a comprehensive and multi-level perception of local details within evolving graphs. In addition, to tackle the issue of unreliable structures in newly added incremental information, the paper introduces an information self-assessment ownership mechanism. This mechanism not only facilitates the propagation of incremental information within the model but also effectively preserves the integrity of past experiences. We design five types of graph memory learning tasks: regular, memory, unlearning, data-incremental, and class-incremental to evaluate BGML. Its excellent performance is confirmed through extensive experiments on multiple node classification datasets.

Handedness and task demands modulate motor cortex lateralization: A cross-sectional fNIRS study.

Risk awareness assessment of construction workers wearing AR head-mounted displays using EEG signals.

This article presents a comparative analysis of five multi-agent architectures based on large, low-generation language models for solving complex legal problems. The study was conducted on a specially prepared dataset of 25 questions of five difficulty levels on Russian family and civil law. Architectures of varying complexity were tested: from a simple lawyer-agent to extended ensembles with a dispatcher and a "jury" system. The main evaluation metrics were the average response quality score, token consumption, economic cost, and efficiency coefficient. The results revealed significant differences between the architectures: Option 5 demonstrated the best quality (6.44 points), but Option 1 proved the most effective with a coefficient of 49.46. Complex architectures required 10-15 times more tokens with an insignificant increase in quality. Analysis by complexity levels revealed that multi-agent systems are most effective for problematic situations and conflicts of laws, while simpler architectures are sufficient for typical tasks. The study provides scientifically based recommendations for selecting optimal architectural solutions for legal advisory systems, balancing quality and cost-effectiveness.

Many researchers nowadays study pronunciation as a means to improve second language students’ communicative skills. Native listeners’ perception tasks are considered a reliable tool although their veracity has been questioned due to the listeners’ subjectivity and/or the type of task performed. The present study investigates the effectiveness of explicit rhythm instruction to Spanish/Catalan bilinguals who were engineering undergraduates to improve their comprehensibility and fluency in English. For this purpose, seven native speakers of English rated the extemporaneous speeches of forty-two students before and after instruction via a 5-point scale. Students were divided into an experimental group, receiving explicit rhythm instruction, and a control group, which did not. Unexpectedly, results showed that the experimental group was less comprehensible after treatment, and that high-level learners of both groups performed worse. After close examination, a ceiling effect was detected, caused by an overrating of students before training. Hence, the use of a mixed-method assessment is encouraged to support human judgments.

Purpose This study aims to enhance construction scheduling through a computational methodology that enables structured input preparation, constraint formulation and combinatorial task analysis to improve schedule feasibility, automation and optimization Design/methodology/approach The proposed methodology integrates building information modeling (BIM), enhanced planning and scheduling (EPS), a structured scheduling approach and constraint programming (CP) to enhance computational and combinatorial scheduling. BIM data is automatically extracted and enriched with material quantities, spatial breakdowns and task types, then structured using EPS into labor-hour–based units and spatial zones. These structured inputs feed into a CP model incorporating precedence logic, resource constraints and execution priorities to generate the construction schedule. Moreover, constraint modification and EPS-driven combinatorial analysis enable alternative scheduling and scenario evaluation. Findings The methodology was applied to a multi-section residential project, resulting in a feasible and optimized construction schedule. The CP model optimized resource use and duration based on the objectives while maintaining logical sequencing, with automated BIM-EPS input reducing manual effort. The schedule was automatically generated based on the predefined constraints and consistency was confirmed using Kendall’s Tau-b correlation. Originality/value This study presents a novel integration of BIM, EPS and CP to advance logic-based and computational scheduling. A key contribution of this study is the implementation of advanced scheduling by facilitating constraint formulation through the EPS methodology and enriched BIM data integration. This approach enables schedule feasibility analysis, improves consistency and addresses limitations of constraint- and logic-based methods through structured input and formalized constraints. By automating the extraction and structuring of data for CP, it reduces some manual effort in data preparation and supports optimization though initial setup and predefined constraints.

ABSTRACT This study investigates how multilingualism and internationalisation intersect to shape students’ experiences of assessment in higher education. Focusing on a trilingual Master’s program, the research draws on semi-structured interviews from a co-research project with students. The analysis identifies three dominant assessment scripts that challenge fairness: assumptions of monolingual proficiency, stable competence across communicative modes, and shared familiarity with academic task types. These scripts often disalign with the diverse linguistic repertoires and educational trajectories of plurilingual students, leading to perceived inequities in assessment. However, the study also highlights pedagogical adjustments – such as integrating students’ profiles, scaffolding the assessment journey, and centering content over form – that students perceive as supportive. Framed through the lense of ‘assessment as cultural script’, the findings call for more reflexive, inclusive assessment practices that acknowledge the linguistic and sociocultural realities of internationalised classrooms. The paper contributes to ongoing debates on equity in higher education and advocates for culturally and linguistically responsive assessment design.

ABSTRACT Computer-supported collaborative concept mapping (CSCCM) integrates concept mapping with technology to support collaborative group learning and conceptual understanding. This study employed a within-subjects experimental design to examine how task type influences learners’ knowledge understanding and behavioral patterns in CSCCM activities. Thirty-four third-year undergraduates completed three tasks – analyze, evaluate, and create – in a CSCCM environment. The results reveal that (1) regarding knowledge understanding, the number of crosslinks per learner in the analyze task was significantly greater than that in the create task; (2) for operational behavioral patterns, learners displayed a distinctive transition sequence, DL (deleting links) → AN (adding nodes), in the analyze task that did not occur in the evaluate or create tasks; and (3) for conversational behavioral patterns, learners engaged in more cognitive-conflict dialog during the create task than during the analyze and evaluate tasks. These findings inform the design of CSCCM activities and the development of adaptive concept mapping systems. Teachers can use collaboration scripts, role assignments, and prompts to reduce cognitive load and enhance engagement. Developers should focus on adaptive tools with real-time suggestions, usability improvements, and automated feedback to refine learners' understanding.

The aim of this study is to identify changes in geographic literacy over time by comparing the success rates of geography applicants when solving identical tasks used in the geography entrance exam test in 2016 and 2024 at the Faculty of Science of Charles University. We summarize the results of a comparison of the success rates in selected test tasks and we look at the shift in geographic literacy between two groups of respondents with similar characteristics (age, interest in studying geography at the same university) in different time periods. Through our study, we aim to open up the possibility of using geography entrance exam tests as one of the possible sources for studying the evolution of geographic literacy over time. Longitudinal studies focusing on changes in geographic literacy are still very rare, which we consider to be a research gap. The tasks used in the entrance tests in 2016 and 2024 were compared in order to determine which types of test tasks experienced the greatest change in success rate. The comparison included answers to a total of 25 test tasks, which were intentionally set identically in both years to make such a comparison possible. Answers were available from 269 respondents in 2016 and from 132 respondents in 2024. When evaluating the results, the tasks were divided according to various criteria (thematic focus, category of educational objectives, use of mathematical skills, inclusion of a visual element, etc.). The results indicate a relatively high rate of change in success in solving certain types and groups of test tasks. The results also show changes in the level of geographical literacy of students who come to university from secondary schools. This information could be helpful not only for universities themselves (who will get better information about changes in the level of applicants from secondary school level), but also for secondary school educators and experts engaged in curriculum development (who will get feedback on secondary education results). The results underline the importance of systematically monitoring changes in geographical literacy and call for further research on a larger dataset and across more time points.

This paper presents a detailed analysis of signal data acquired from wearable sensors such as accelerometers and barometric altimeters for human activity recognition, with an emphasis on fall detection. This research addressed two types of activity recognition tasks: a binary classification problem between activities of daily living (ADLs) and simulated fall activities and a multiclass classification problem involving five different activities (running, walking, sitting down, jumping, and falling). By combining features derived from both sensors, traditional machine models such as random forest, support vector machine, XGBoost, logistic regression, and majority voter models were used for both classification problems. All of the aforementioned methods generally produced better results using combined features of both sensors compared to single-sensor models, highlighting the potential of sensor fusion approaches for fall detection.