doi
stringlengths 17
26
| title
stringlengths 22
300
| abstract
stringlengths 61
19.2k
⌀ | journal_eissn
stringclasses 12
values | journal_category
stringclasses 4
values |
|---|---|---|---|---|
10.3389/frai.2025.1540646
|
InGSA: integrating generalized self-attention in CNN for Alzheimer's disease classification
|
Alzheimer's disease (AD) is an incurable neurodegenerative disorder that slowly impair the mental abilities. Early diagnosis, nevertheless, can greatly reduce the symptoms that are associated with the condition. Earlier techniques of diagnosing the AD from the MRI scans have been adopted by traditional machine learning technologies. However, such traditional methods involve depending on feature extraction that is usually complex, time-consuming, and requiring substantial effort from the medical personnel. Furthermore, these methods are usually not very specific as far as diagnosis is concerned. In general, traditional convolutional neural network (CNN) architectures have a problem with identifying AD. To this end, the developed framework consists of a new contrast enhancement approach, named haze-reduced local-global (HRLG). For multiclass AD classification, we introduce a global CNN-transformer model InGSA. The proposed InGSA is based on the InceptionV3 model which is pre-trained, and it encompasses an additional generalized self-attention (GSA) block at top of the network. This GSA module is capable of capturing the interaction not only in terms of the spatial relations within the feature space but also over the channel dimension it is capable of picking up fine detailing of the AD information while suppressing the noise. Furthermore, several GSA heads are used to exploit other dependency structures of global features as well. Our evaluation of InGSA on a two benchmark dataset, using various pre-trained networks, demonstrates the GSA's superior performance.
|
26248212
|
AI
|
10.3390/ai6030058
|
Leveraging Spectral Neighborhood Information for Corn Yield Prediction with Spatial-Lagged Machine Learning Modeling: Can Neighborhood Information Outperform Vegetation Indices?
|
Accurate and reliable crop yield prediction is essential for optimizing agricultural management, resource allocation, and decision-making, while also supporting farmers and stakeholders in adapting to climate change and increasing global demand. This study introduces an innovative approach to crop yield prediction by incorporating spatially lagged spectral data (SLSD) through the spatial-lagged machine learning (SLML) model, an enhanced version of the spatial lag X (SLX) model. The research aims to show that SLSD improves prediction compared to traditional vegetation index (VI)-based methods. Conducted on a 19-hectare cornfield at the ARS Grassland, Soil, and Water Research Laboratory during the 2023 growing season, this study used five-band multispectral image data and 8581 yield measurements ranging from 1.69 to 15.86 Mg/Ha. Four predictor sets were evaluated: Set 1 (spectral bands), Set 2 (spectral bands + neighborhood data), Set 3 (spectral bands + VIs), and Set 4 (spectral bands + top VIs + neighborhood data). These were evaluated using the SLX model and four decision-tree-based SLML models (RF, XGB, ET, GBR), with performance assessed using R2 and RMSE. Results showed that incorporating spatial neighborhood data (Set 2) outperformed VI-based approaches (Set 3), emphasizing the importance of spatial context. SLML models, particularly XGB, RF, and ET, performed best with 4–8 neighbors, while excessive neighbors slightly reduced accuracy. In Set 3, VIs improved predictions, but a smaller subset (10–15 indices) was sufficient for optimal yield prediction. Set 4 showed slight gains over Sets 2 and 3, with XGB and RF achieving the highest R2 values. Key predictors included spatially lagged spectral bands (e.g., Green_lag, NIR_lag, RedEdge_lag) and VIs (e.g., CREI, GCI, NCPI, ARI, CCCI), highlighting the value of integrating neighborhood data for improved corn yield prediction. This study underscores the importance of spatial context in corn yield prediction and lays the foundation for future research across diverse agricultural settings, focusing on optimizing neighborhood size, integrating spatial and spectral data, and refining spatial dependencies through localized search algorithms.
|
26732688
|
AI
|
10.3390/ai6030059
|
Clinical Applicability of Machine Learning Models for Binary and Multi-Class Electrocardiogram Classification
|
Background: This study investigates the application of machine learning models to classify electrocardiogram signals, addressing challenges such as class imbalances and inter-class overlap. In this study, “normal” and “abnormal” refer to electrocardiogram findings that either align with or deviate from a standard electrocardiogram, warranting further evaluation. “Borderline” indicates an electrocardiogram that requires additional assessment to distinguish benign variations from pathology. Methods: A hierarchical framework reformulated the multi-class problem into two binary classification tasks—distinguishing “Abnormal” from “Non-Abnormal” and “Normal” from “Non-Normal”—to enhance performance and interpretability. Convolutional neural networks, deep neural networks, and tree-based models, including Gradient Boosting Classifier and Random Forest, were trained and evaluated using standard metrics (accuracy, precision, recall, and F1 score) and learning curve convergence analysis. Results: Results showed that convolutional neural networks achieved the best balance between generalization and performance, effectively adapting to unseen data and variations without overfitting. They exhibit strong convergence and robust feature importance rankings, with ventricular rate, QRS duration, and P-R interval identified as key predictors. Tree-based models, despite their high performance metrics, demonstrated poor convergence, raising concerns about their reliability on unseen data. Deep neural networks achieved high sensitivity but suffered from overfitting, limiting their generalizability. Conclusions: The hierarchical binary classification approach demonstrated clinical relevance, enabling nuanced diagnostic insights. Furthermore, the study emphasizes the critical role of learning curve analysis in evaluating model reliability, beyond performance metrics alone. Future work should focus on optimizing model convergence and exploring hybrid approaches to improve clinical applicability in electrocardiogram signal classification.
|
26732688
|
AI
|
10.1007/s00432-025-06157-2
|
Elevated platelet distribution width and diabetes may serve as preoperative predictors of microvascular invasion in primary hepatocellular carcinoma
|
Background and objective Hepatocellular carcinoma (HCC) is one of the malignancies with increasing incidence globally, and microvascular invasion (MVI) is a crucial determinant of prognosis in patients. This study aimed to investigate platelet distribution width (PDW) and diabetes mellitus as indicators for predicting preoperative MVI in HCC, providing more accurate predictive tools for clinicians to guide treatment strategies and improve patient survival and quality of life. Methods A retrospective study was conducted, including 1357 patients who underwent hepatectomy for HCC between January 2008 and December 2014 at the Eastern Hepatobiliary Surgery Hospital in China. Clinical, pathological, and radiological data, including PDW and diabetes status, were collected. Univariate and multivariate logistic regression analyses were performed to identify risk factors for MVI and establish a predictive model. The predictive performance of the model was evaluated through nomograms and internal validation. Results Univariate analysis revealed significant associations between MVI and diabetes mellitus, presence of liver cirrhosis, prealbumin level, tumor diameter, number of tumors, HBV DNA viral load > 104, and PDW ≥ 17. Multivariate logistic regression analysis identified diabetes mellitus, liver cirrhosis, prealbumin level, tumor diameter, number of tumors, HBV DNA viral load > 104, and PDW ≥ 17 as independent risk factors for MVI. Based on these findings, a predictive model was established, demonstrating high predictive accuracy and stability in both the training and validation cohorts. Conclusion This study confirmed PDW and diabetes mellitus as reliable indicators for predicting preoperative MVI in HCC and established a corresponding predictive model. Future research should further explore the underlying mechanisms and enhance clinical validation to advance the field of HCC treatment.
|
14321335
|
ONCOLOGY
|
10.1186/s40359-025-02581-8
|
Psychometric properties of the Persian version of the ambivalent ageism scale (benevolent and hostile) in the adult population in Iran
|
With the growing population of older adults and the prevalence of negative attitudes towards them, the issue of ageism and its health and economic impacts in both benevolent and hostile contexts warrants special attention. It is crucial to examine the attitudes of other age groups towards older adults across different societies. Particularly, the benevolent dimension of ageism, which has been less explored in research, requires more focus. Therefore, this study aims to conduct a psychometric evaluation of The Ambivalent Ageism Scale among the adult population in Iran. This methodological study was conducted in comprehensive health centers in Gorgan city in 2023. A total of 381 eligible adults participated. The Ambivalent Ageism Scale (AAS) was utilized, and the psychometric assessment included translation, face validity, and content validity. Additionally, exploratory factor analysis and confirmatory factor analysis were performed. The reliability of the scale was evaluated using the internal consistency method. The research findings were analyzed using SPSS and AMOS software version 24. Qualitative face and content validity assessments led to textual and editorial modifications of the items. The Content Validity Ratio (CVR), Item-Content Validity Index (I-CVI), and Kappa (K*) scores were acceptable for all items. In the exploratory factor analysis (EFA), similar to the original questionnaire, three factors were extracted, accounting for approximately 54% of the total variance. The fit indices in the confirmatory factor analysis (CFA) indicated an acceptable model fit. During CFA, four items were eliminated. The reliability of the entire questionnaire was deemed acceptable with a Cronbach’s alpha coefficient of 0.763. Consequently, the Persian version of the Ambivalent Ageism Scale was confirmed with nine items. The Persian version of The Ambivalent Ageism Scale demonstrates sufficient validity and reliability for measuring attitudes towards aging within Iranian society. Given the cultural adaptation of this tool, the questionnaire can be utilized to assess adults’ views and attitudes towards older adults in both hostile and benevolent dimensions. Furthermore, it can aid in formulating family-oriented policies for older adult care and facilitate improvements in the quality of care for this population group.
|
20507283
|
PSYCHOLOGY
|
10.3390/ai6030062
|
AI-Driven Telerehabilitation: Benefits and Challenges of a Transformative Healthcare Approach
|
Artificial intelligence (AI) has revolutionized telerehabilitation by integrating machine learning (ML), big data analytics, and real-time feedback to create adaptive, patient-centered care. AI-driven systems enhance telerehabilitation by analyzing patient data to personalize therapy, monitor progress, and suggest adjustments, eliminating the need for constant clinician oversight. The benefits of AI-powered telerehabilitation include increased accessibility, especially for remote or mobility-limited patients, and greater convenience, allowing patients to perform therapies at home. However, challenges persist, such as data privacy risks, the digital divide, and algorithmic bias. Robust encryption protocols, equitable access to technology, and diverse training datasets are critical to addressing these issues. Ethical considerations also arise, emphasizing the need for human oversight and maintaining the therapeutic relationship. AI also aids clinicians by automating administrative tasks and facilitating interdisciplinary collaboration. Innovations like 5G networks, the Internet of Medical Things (IoMT), and robotics further enhance telerehabilitation’s potential. By transforming rehabilitation into a dynamic, engaging, and personalized process, AI and telerehabilitation together represent a paradigm shift in healthcare, promising improved outcomes and broader access for patients worldwide.
|
26732688
|
AI
|
10.1186/s40359-025-02600-8
|
Psychometric properties of the Persian version of the suicidal intrusions attributes scale (SINAS) in patients with suicidal attempt
|
The Suicidal Intrusions Attributes Scale (SINAS) is a brief self-report measure designed to assess the frequency, distress, and controllability of suicidal intrusions—vivid, uncontrollable mental images and thoughts related to suicide or its aftermath. Despite its clinical relevance, its psychometric properties remain underexplored. This study aimed to evaluate the psychometric properties of the Persian version of the SINAS. A cross-sectional design was employed. 304 outpatients (aged 18 to 65, M = 27.27, SD = 8.53) including 243 males and 61 females with a history of suicide attempts were recruited using a convenience sampling method from psychiatric clinics and hospitals in Tehran. Participants completed the SINAS along with the Beck Depression Inventory-II (BDI-II) to assess depressive symptoms, the Beck Hopelessness Scale (BHS) to measure negative expectations about the future, the Beck Scale for Suicide Ideation (BSSI) to evaluate suicidal thoughts and intentions, and the Suicide Behaviors Questionnaire-Revised (SBQ-R) to assess past suicidal behaviors and future risk. Confirmatory factor analysis supported a one-factor structure of the SINAS, which was invariant across gender groups. The scale demonstrated strong internal consistency and good test-retest reliability over a two-week interval. Additionally, the SINAS showed significant associations with depressive symptoms, hopelessness, suicide ideation, and suicide risk behaviors, supporting its convergent validity. Overall, the findings indicate that the Persian version of the SINAS is a valid and reliable instrument for assessing suicidal intrusions in both clinical and research settings in Iran.
|
20507283
|
PSYCHOLOGY
|
10.1007/s44196-025-00781-z
|
A Deep Learning Model Leveraging Time-Series System Call Data to Detect Malware Attacks in Virtual Machines
|
A Tenant Virtual Machine (TVM) user in the cloud may misuse its computing power to launch malware attack against other tenant VMs, Host OS, Hypervisor, or any other computing devices/resources inside the cloud environment of a Cloud Service Provider. The security solutions deployed within the TVM may not be reliable, as malware can disable them or remain undetected due to its hidden nature. Therefore, security solutions deployed outside the virtual machine are necessary. This research proposes deploying an Intrusion Detection System (IDS) at the Hypervisor layer, utilizing time series system call data and employing a Convolutional Neural Network (CNN) model to accurately detect the presence of malicious (malware) computer programs within virtual machines. The raw VMM system call traces are transformed into novel Time Series System Call patterns and utilized by a deep learning algorithm for training and building the classifier model. A deep learning model, CNN, is used to build the classifier model for detecting intrusions with high accuracy. It is capable of detecting both known and unknown malware. The CNN model is compared with machine learning algorithms for the results and discussions, and it outperforms ML algorithms in terms of intrusion detection accuracy when utilizing novel time series system call data..
|
18756883
|
AI
|
10.3390/ejihpe15030037
|
Linguistic and Cognitive Abilities in Children with Dyslexia: A Comparative Analysis
|
Introduction: Dyslexia is a prevalent learning disorder that significantly affects the child population. It is often accompanied by deficits in language processes, cognition, and executive functioning, all of which are crucial for reading development. Children with dyslexia frequently exhibit difficulties in phonological processing, semantics, morphosyntax, and also in cognitive areas such as working memory, inhibition, planning, and attention. Objective: The primary objective of this study was to compare the linguistic, cognitive, and executive functioning abilities between children diagnosed with dyslexia and those with typical reading development. Methodology: A total of 120 children were selected and divided into two groups: the G-DYSLEXIA group (n = 60), consisting of children diagnosed with dyslexia, and the G-CONTROL group (n = 60), with typical reading development. Language, cognition, and executive functions were assessed using standardized tests: CELF-5, WISC-V, and ENFEN. Statistical analyses included descriptive statistics, independent sample t-tests, and Chi-square tests to compare the performance between these two groups. Results: The study revealed significant differences between the two groups in all dimensions assessed. Specifically, children with dyslexia showed markedly lower performance in linguistic, cognitive, and executive functioning measures compared with their peers with typical development. Conclusion: Children with dyslexia present a distinct clinical profile characterized by significant difficulties in language processing, cognition, and executive functions. These challenges interfere with their reading acquisition and academic performance, limiting their integration into educational environments and impacting their overall quality of life.
|
22549625
|
PSYCHOLOGY
|
10.3390/educsci15030375
|
Technology-Enhanced Language Learning: Subtitling as a Technique to Foster Proficiency and Intercultural Awareness
|
Computer-Assisted Language Learning (CALL) is an umbrella term that encompasses diverse technologies with the purpose of enhancing language learning. In the existing literature on CALL, intercultural awareness and the pedagogical use of multimedia products have received less attention. This study explores how the process of creating subtitles for short clips may enhance language skills and intercultural awareness when implemented through lesson plans designed following the framework proposed by the TRADILEX project. A pre-experimental longitudinal design was implemented. The sample consisted of 43 participants who were enrolled in a B2 English course at the University of Córdoba (Spain). During the course, participants consistently attended theoretical sessions. The intervention took place during the practical sessions from February to April 2024, involving four subtitling-based lesson plans on literature and gender. After the intervention, the practical sessions shifted to a traditional, textbook-based format from April to June 2024. The instruments employed to assess the effectiveness of the intervention consisted of a commercial test by MacMillan and the ERI scale on interculturality. The results showed that after the intervention, there was a significant improvement in language proficiency, which increased at a slower rate during the traditional sessions. However, when it comes to intercultural awareness, there was a peak of attainment after the intervention, but attrition rapidly took place. Regarding the pedagogical implications of this study, subtitling could be an appropriate technique that allows contact with the L2 culture and shows positive effects in terms of proficiency.
|
22277102
|
EDUCATION
|
10.3390/ejihpe15030039
|
The Moving Mandala: Exploring the Pro-Social Effects of Musical and Non-Musical Synchrony in Children in a Virtual World
|
Synchronous movement between individuals has been shown to increase pro-sociality, such as closeness and generosity. To date, synchrony research tests these effects using a variety of movement tasks, including musical and non-musical coordination. However, musical versus non-musical synchrony may have separable pro-social effects. To test this, we had 60 children immersed in an augmented reality space called the ‘Moving Mandala’ where they moved asynchronously with only visual cues, synchronously with only visual cues or synchronously with musical and visual cues. We then tested for differences in pro-social effects using sharing and proxemics tasks. Results showed that while the synchrony version of the mandala led to greater closeness in the proxemics task, the musical synchrony led to more pro-sociality on the sharing task. The implications of these findings are discussed.
|
22549625
|
PSYCHOLOGY
|
10.3390/cancers17061021
|
Correction: Zossou et al. Radiomics-Based Classification of Tumor and Healthy Liver on Computed Tomography Images. Cancers 2024, 16, 1158
| null |
20726694
|
ONCOLOGY
|
10.3390/ejihpe15030041
|
Risk Assessment Profiles for Caregiver Burden in Family Caregivers of Persons Living with Alzheimer’s Disease: An Exploratory Study with Machine Learning
|
Alzheimer’s disease (AD) places a profound global challenge, driven by its escalating prevalence and the multifaceted strain it places on individuals, families, and societies. Family caregivers (FCs), who are pivotal in supporting family members with AD, frequently endure substantial emotional, physical, and psychological demands. To better understand the determinants of family caregiving strain, this study employed machine learning (ML) to develop predictive models identifying factors that contribute to caregiver burden over time. Participants were evaluated across sociodemographic clinical, psychophysiological, and psychological domains at baseline (T1; N = 130), six months (T2; N = 114), and twelve months (T3; N = 92). Results revealed three distinct risk profiles, with the first focusing on T2 data, highlighting the importance of distress, forgiveness, age, and heart rate variability. The second profile integrated T1 and T2 data, emphasizing additional factors like family stress. The third profile combined T1 and T2 data with sociodemographic and clinical features, underscoring the importance of both assessment moments on distress at T2 and forgiveness at T1 and T2, as well as family stress at T1. By employing computational methods, this research uncovers nuanced patterns in caregiver burden that conventional statistical approaches might overlook. Key drivers include psychological factors (distress, forgiveness), physiological markers (heart rate variability), contextual stressors (familial dynamics, sociodemographic disparities). The insights revealed enable early identification of FCs at higher risk of burden, paving the way for personalized interventions. Such strategies are urgently needed as AD rates rise globally, underscoring the imperative to safeguard both patients and the caregivers who support them.
|
22549625
|
PSYCHOLOGY
|
10.3390/educsci15030395
|
Creativity and Preservice Teachers: A Literature Review of an Underexplored Field (2014–2024)
|
This systematic literature review examines the relationship between creativity and preservice teachers in scientific publications from 2014 to 2024. Using the PRISMA methodology, 27 empirical articles were selected based on their relevance to the research focus. The study provides both a bibliometric overview of the field and a substantive analysis of existing knowledge. Key findings reveal significant dispersion within the field, a proliferation of diverse definitions of creativity, and limited attention to the specific characteristics of preservice teachers in the research. Four central themes emerged: beliefs about creativity, personal characteristics, the creative processes, and teaching for creativity. These themes highlight the fragmented yet evolving nature of the discourse. The paper underscores the necessity of more comprehensive research approaches that transcend methodological individualism and better capture the domain-specific nature of creativity in preservice teachers. By integrating these perspectives, the study aims to advance a more cohesive understanding of how creativity can be cultivated in teacher preparation.
|
22277102
|
EDUCATION
|
10.3390/ai6040063
|
FedBirdAg: A Low-Energy Federated Learning Platform for Bird Detection with Wireless Smart Cameras in Agriculture 4.0
|
Birds can cause substantial damage to crops, directly affecting farmers’ productivity and profitability. As a result, detecting bird presence in crop fields is crucial for effective crop management. Traditional agricultural practices have used various tools and techniques to deter pest birds, while digital agriculture has advanced these efforts through Internet of Things (IoT) and artificial intelligence (AI) technologies. With recent advancements in hardware and processing chips, connected devices can now utilize deep convolutional neural networks (CNNs) for on-field image classification. However, training these models can be energy-intensive, especially when large amounts of data, such as images, need to be transmitted for centralized model training. Federated learning (FL) offers a solution by enabling local training on edge devices, reducing data transmission costs and energy demands while also preserving data privacy and achieving shared model knowledge across connected devices. This paper proposes a low-energy federated learning framework for a compact smart camera network designed to perform simple image classification for bird detection in crop fields. The results demonstrate that this decentralized approach achieves performance comparable to a centrally trained model while consuming at least 8 times less energy. Further efficiency improvements, with a minimal tradeoff in performance reduction, are explored through early stopping.
|
26732688
|
AI
|
10.3390/cancers17071079
|
Adenoid Cystic Carcinoma of the Breast: A Narrative Review and Update on Management
|
Rare breast malignancies represent a challenge for treatment decision making given the lack of evidence-based guidelines. As a particularly uncommon tumor, adenoid cystic carcinomas are especially challenging. Although, histopathologically, they share the same tumor molecular profile as hormone receptor-negative and HER2 nonamplified carcinomas with aggressive physiology, adenoid cystic carcinomas generally have a favorable prognosis. Thus, there is evidence to suggest that more aggressive treatment regimens may not provide better therapeutic effects. In this review, we discuss ACCB with the goal of highlighting pathophysiology, clinical features, and treatment strategies. Existing data support consideration for adjuvant radiation with the omission of axillary staging and systemic therapies.
|
20726694
|
ONCOLOGY
|
10.1186/s40359-025-02578-3
|
A socially prescribed creative play intervention for new parents: investigating post traumatic stress around birth and changes in postnatal depression and reflective function
|
Background: Parenthood is a key transition period which involve emotional, social and physical adjustments. Social prescribing is a method that connects people to community-based activities, groups, and services to addressing various needs impacting their health and wellbeing. This pilot investigation aimed to assess whether a curated socially prescribed creative play programme would impact upon new parents’ social connection, mental health and reflective function through a programme designed to support these changes. Methods: This study was part of a 5-week long socially prescribed creative play programme at a family theatre company in the North of England, aimed at providing social capital to families while teaching creative play. In total, 57 parents (M = 30.73, SD = 6.20) completed baseline and post-intervention measures of birth trauma experiences (City Birth Trauma Scale), postnatal depression (Edinburgh Postnatal Depression Scale) reflective function (Reflective Functioning Questionnaire), and qualitative, open-ended questions on social opportunities. Descriptive analyses were completed using t-tests and chi-square tests, while repeated measures ANOVAs were used to answer questions around the main analyses. Results: The participants experienced a statistically significant reduction in postnatal depression scores following the intervention, but no changes were found in reflective function or birth trauma scores; secondly, birth trauma scores predicted later depression scores as well as reflective functioning uncertainty scores (but not certainty scores). Qualitative analysis found social opportunities were not why parents came but was, after attending, their favourite part of the socially prescribed programme. Those parents reporting on social opportunities were more likely to reference their own needs while non-social activities were associated with their child’s needs. Conclusions: Socially prescribed creative play programmes for new parents could be a “waiting well” intervention. A longer duration and trauma informed focus would need to be considered in future cohorts.
|
20507283
|
PSYCHOLOGY
|
10.3390/ai6040064
|
SMART Restaurant ReCommender: A Context-Aware Restaurant Recommendation Engine
|
With the rise of e-commerce and web application usage, recommendation systems have become important to our daily tasks. They provide personalized suggestions to assist with any task under consideration. While various machine learning algorithms have been developed for recommendation tasks, existing systems still face limitations. This research focuses on advancing context-aware recommendation sytems by leveraging the capabilities of Large Language Models (LLMs) in conjunction with real-time data. The research exploits the integration of existing real-time data APIs with LLMs to enhance the capabilities of the recommendation systems already integrated into smart societies. The experimental results demonstrate that the hybrid approach significantly improves the user experience and recommendation quality, ensuring more relevant and dynamic suggestions.
|
26732688
|
AI
|
10.3389/feduc.2025.1541475
|
Adapting to crisis and unveiling the digital shift: a systematic literature review of digital competence in education related to COVID-19
|
Nowadays, with technology penetrating into every aspect of our life, the ways to acquire knowledge has been greatly revolutionized. The outbreak of the Coronavirus (COVID-19) has accelerated digital informatization in education and the educational model has been transformed substantially. The demand for digital competence is at record high. The purpose of this study is to systematically explore digital competence in different national educational contexts during the COVID-19 pandemic (2019–2021), to provide academics with the current state of digital competence in education and main research trends in digital competence in education during this period, elucidate the impact of pandemic on digital competence, and explore the limitations in the implementation of digital competence in educational research. The results indicate that most research on digital competence in educational contexts related to COVID-19 focused on the current state of the digital competence of teachers and students, especially those in higher education and formal learning context. Still, with the situation compounded, the researchers furthered their study by investigating the factors that influenced digital competence in order to address educational challenges in a pandemic context. In addition, teachers and students were still not well equipped as for digital competence though their digital awareness and digital readiness in the teaching and learning process increased. It is recommended to promote and enhance digital competence training in order to improve students’ achievement and the quality of education.
|
2504284X
|
EDUCATION
|
10.3390/educsci15040415
|
Client Violence Against Educational Workers: A Systematic Review
|
Client-initiated workplace violence in educational settings is a global issue affecting both teaching and non-teaching employees, such as instructional assistants, counselors, and administrators, among other school workers. Although studies on violence in educational settings have primarily focused on students, there has been growing interest in examining violence against teachers and, more recently, against teaching assistants and other educational professionals. This systematic review aims to analyze studies from diverse educational settings to examine the characteristics, causes, effects, and coping strategies associated with violence perpetrated by students, parents, or guardians, with the goal of informing and advancing prevention strategies. Following the PRISMA 2020 guidelines, a systematic literature review was conducted, analyzing studies across various educational environments to examine the characteristics, causes, effects, and coping strategies of violence perpetrated by students, parents, or guardians. This review revealed a significant prevalence of physical, psychological, and verbal assaults. However, most studies originated from Anglo-Saxon contexts, limiting their generalizability to diverse cultural and educational settings. The lack of research in other languages and in underrepresented regions highlights critical gaps in understanding this issue globally. The revision conclude that workplace violence in educational settings demands urgent and comprehensive responses involving all stakeholders. Implementing targeted prevention strategies and fostering a culture of respect are essential to ensure safe and healthy learning environments.
|
22277102
|
EDUCATION
|
10.3390/ai6040066
|
One-Shot Autoregressive Generation of Combinatorial Optimization Solutions Based on the Large Language Model Architecture and Learning Algorithms
|
Large Language Models (LLMs) have immensely advanced the field of Artificial Intelligence (AI), with recent models being able to perform chain-of-thought reasoning and solve complex mathematical problems, ranging from theorem proving to ones involving advanced calculus. The success of LLMs derives from a combination of the Transformer architecture with its attention mechanism, the autoregressive training methodology with masked attention, and the alignment fine-tuning via reinforcement learning algorithms. In this research, we attempt to explore a possible solution to the fundamental NP-hard problem of combinatorial optimization, in particular, the Traveling Salesman Problem (TSP), by following the LLM approach in terms of the architecture and training algorithms. Similar to the LLM design, which is trained in an autoregressive manner to predict the next token, our model is trained to predict the next node in a TSP graph. After the model is trained on random TSP graphs with known near-optimal solutions, we fine-tune the model using Direct Preference Optimization (DPO). The tour generation in a trained model is autoregressive one-step generation with no need for iterative refinement. Our results are very promising and indicate that, for TSP graphs up to 100 nodes, a relatively small amount of training data yield solutions within a few percent of the optimal. This optimization improves if more data are used to train the model.
|
26732688
|
AI
|
10.3390/ejihpe15040048
|
Understanding the Use of Social and Emotional Learning in Elementary Schools: A Theory of Planned Behaviour Perspective
|
Research has demonstrated that social–emotional learning (SEL) positively influences myriad domains of children’s development. However, the underlying mechanisms influencing teachers’ adoption of SEL remain underexplored. Guided by the Theory of Planned Behaviour (TPB), this quantitative cross-sectional study sought to elucidate the factors that motivate teachers to adopt SEL teaching practices. The study’s sample included 166 volunteer teachers in Luxembourg, recruited as part of a nationwide educational survey. Of these, 82.5% were women. Participants were recruited through convenience sampling, ensuring diversity in socio-economic backgrounds, grade levels, and student needs. Although these findings are based on self-reported data, they offer novel insights by quantifying teachers’ engagement with SEL, with over 50% already implementing related activities. Structural equation modelling shows that the TPB model accounted for 49% of the variance in teachers’ intentions and 44% of the variance in the adoption of SEL practices. Higher intention and self-efficacy predicted more frequent SEL implementation. Teachers with positive SEL attitudes and higher self-efficacy showed greater intention to implement SEL. These findings underscore the significance of cultivating positive attitudes and self-efficacy to facilitate the effective implementation of SEL in educational settings. The role of teacher gender and audience, as well as implications for teaching, professional development, and SEL research, are discussed.
|
22549625
|
PSYCHOLOGY
|
10.3389/frai.2025.1558938
|
Enhancing structured data generation with GPT-4o evaluating prompt efficiency across prompt styles
|
Large language models (LLMs), such as GPT-4o, provide versatile techniques for generating and formatting structured data. However, prompt style plays a critical role in determining the accuracy, efficiency, and token cost of the generated outputs. This paper explores the effectiveness of three specific prompt styles–JSON, YAML, and Hybrid CSV/Prefix–for structured data generation across diverse applications. We focus on scenarios such as personal stories, receipts, and medical records, using randomized datasets to evaluate each prompt style's impact. Our analysis examines these prompt styles across three key metrics: accuracy in preserving data attributes, token cost associated with output generation, and processing time required for completion. By incorporating structured validation and comparative analysis, we ensure precise evaluation of each prompt style's performance. Results are visualized through metrics-based comparisons, such as Prompt Style vs. Accuracy, Prompt Style vs. Token Cost, and Prompt Style vs. Processing Time. Our findings reveal trade-offs between prompt style complexity and performance, with JSON providing high accuracy for complex data, YAML offering a balance between readability and efficiency, and Hybrid CSV/Prefix excelling in token and time efficiency for flat data structures. This paper explores the pros and cons of applying the GPT-4o LLM to generate structured data. It also provides practical recommendations for selecting prompt styles tailored to specific requirements, such as data integrity, cost-effectiveness, and real-time processing needs. Our findings contribute to research on how prompt engineering can optimize structured data generation for AI-driven applications, as well as documenting limitations that motivate future work needed to improve LLMs for complex tasks.
|
26248212
|
AI
|
10.1186/s40359-025-02620-4
|
Exploring the factors influencing the adoption of artificial intelligence technology by university teachers: the mediating role of confidence and AI readiness
|
This study aims to explore the mediating role of confidence and artificial intelligence (AI) readiness in university teachers’ behavioral intention to adopt AI technology, providing empirical support for enhancing teachers’ willingness to use AI technology from both theoretical and practical perspectives. This study used a random sampling method to conduct an online survey of 504 university teachers, assessing the impact of subjective norms on behavioral intention. The survey included scales for subjective norms, confidence, AI readiness, and behavioral intention. Data analysis was performed using AMOS 26, SPSS Statistics 27 software and Model 6 from the PROCESS 4.0 plugin, aiming to investigate the mediating role of confidence and AI readiness between subjective norms and behavioral intention. Subjective norms were found to have a significant positive correlation with behavioral intention. Subjective norms indirectly influenced behavioral intention through confidence or AI readiness. Confidence and AI readiness played a chain-mediating role in the relationship between subjective norms and behavioral intention (β = 0.0324, 95% CI: [0.0129, 0.0551]), accounting for 12.87% of the total effect. This study reveals the positive role of subjective norms in university teachers’ behavioral intention to adopt AI technology, indicating that subjective norms not only directly enhance behavioral intention but also exert indirect effects through both single and chain mediation of confidence and AI readiness. The findings highlight the critical role of confidence and AI readiness in the relationship between subjective norms and behavioral intention, suggesting that to effectively increase university teachers’ willingness to use AI technology, it is important to focus on improving their confidence in and readiness for AI technology, thereby strengthening the positive impact of subjective norms.
|
20507283
|
PSYCHOLOGY
|
10.3390/ai6040068
|
Voice-AttentionNet: Voice-Based Multi-Disease Detection with Lightweight Attention-Based Temporal Convolutional Neural Network
|
Voice data contain a wealth of temporal and spectral information and can be a valuable resource for disease classification. However, traditional methods are often not effective in capturing the key features required for the classification of multiple disease classes. To address this challenge, we propose a voice-based multi-disease detection approach with a lightweight attention-based temporal convolution neural network (Voice-AttentionNet) designed to analyze speech data for multi-class disease classification. Our model utilizes the temporal convolution neural network (CNN) architecture to extract high-resolution temporal features, while incorporating attention mechanisms to highlight disease-related patterns. Extensive experiments have been conducted on our dataset, including speech samples from patients with multiple illnesses. The results show that our method achieves the most advanced performance with an average classification accuracy of 91.61% on six datasets and is superior to the existing classical models. These findings highlight the potential of combining attention mechanisms with temporal CNNs in the use of speech data for disease classification. Moreover, this study provides a promising direction for deploying AI-driven diagnostic tools in clinical scenarios.
|
26732688
|
AI
|
10.3389/frai.2025.1568210
|
AI in business operations: driving urban growth and societal sustainability
|
Approximately 30% of smart city applications will use artificial intelligence (AI) by the end of 2025, thereby radically altering the urban sustainability landscape in the future (Yan et al., 2023). The advent of AI in reshaping traditional businesses into sustainable operations is evident. Whenever AI is brought to the forefront, it is considered a cornerstone in the business domain, enabling a transition towards more innovative and sustainable practices (Appio et al., 2024). Incorporating AI into business practices has many facets. According to Grand View Research (2023), the global AI market size was anticipated at USD 196.63 billion in 2023 and is expected to grow at a CAGR of 36.6% from 2024 to 2030. The recent fanfare surrounding AI has elevated it to a key enabler of sustainable development, prompting many companies to prioritize and integrate it into their business operations; hence, there is a stark difference between traditional and new practices. In tandem with this evolution, urban growth and societal dynamics are experiencing profound changes as AI-driven solutions come to the fore in various aspects of modern society (Shahidi Hamedani et al., 2024). AI applications in city government, transforming conventional cities into efficient ones (Ortega-Fernández et al., 2020), have significantly shifted from functional systems to more sustainable and intelligent ones. Furthermore, from another perspective, the role of AI in optimizing business processes has surpassed comparison with its implication for improving logistics operational capabilities and reducing environmental impacts (Jorzik et al., 2024a) till manufacturing reduces downtime, all of which contribute to the growth of urban economics. In the meantime, with the speedy pace of adoption of AI in business operations, it is also imperative to amalgamate with sustainable practices. Acting on this matter requires a thoughtful approach that aligns AI with social, economic, and environmental sustainability.The intersection of AI role and business operations has recently gained widespread attention. Some studies (Chen et al., 2024;Shahzadi et al., 2024)focused on AI's role in supply chain management, highlighting its role in minimizing inefficiencies and improving logistics by utilizing AI more often;supply chains become leaner and reduced carbon footprints, paving the path to sustainable operations. It is estimated that by 2026, 60% of businesses will adopt AI-powered warehouse solutions instead of just 10% in 2020 (MHI, 2024).In line with this shift, (Dilmegani & Ermut, 2025) note that businesses also invest heavily in warehouse robots to enhance their supply chain management through AI technology. Robots can manage operations more efficiently and accurately by automating picking, packing, sorting, and inventory management, thus saving labor costs and accelerating order processing. Amazon, for instance, has deployed more than 200,000 robots in its warehouses to optimize operations.AI can be used to optimize resource utilization, automate processes for improved efficiency, and enable real-time monitoring that aligns with sustainability goals (Waltersmann et al., 2021). As sustainable supply chain management focuses on reducing waste and enhancing traceability, AI-driven technologies such as machine learning and big data analytics have been pivotal in achieving these goals. (Tsolakis et al., 2023) Companies like eBay leverage AI for machine translation, enhancing decision-making and operational efficiency . Similarly, Vodafone employs AI-driven analytics to personalize services, exemplifying its transformative impact. (Jorzik et al., 2024a).These technologies help reduce forecasting errors, minimize excess inventory, and lower energy consumption. (Sharma et al., 2020) Likewise, Smart grid protection sensors can detect defects up to 80% more accurately than traditional sensors, reducing losses and improving the system's reliability by adjusting to grid conditions dynamically (Mahadik, Sheetal et al., 2025). These applications contribute to urban economic growth by fostering technological innovation. AI leverages advanced techniques like deep reinforcement learning (DRL) to optimize dynamic business operations (Shuford, 2024). DRL improves supply chain management through adaptive routing and inventory optimization, dynamically adjusting to real-time changes in demand and logistics; with the help of DRL, researchers can develop systems that can dynamically adapt to changes, optimize resource utilization, and facilitate multi-objective decision-making for instance, (Dehaybe et al., 2024).In addition, it enables businesses to prevent equipment failures and minimize downtime, thereby streamlining workflows significantly (Mohan et al., 2021). Moreover, in urban centers, these advancements catalyze economic growth and foster innovation. In other words, a key contribution of AI is to facilitate smart urban development and efficient resource allocation, thereby ensuring that cities are resilient and economically prosperous (Li et al., 2024). In developing smart cities, AI has a transformative impact on urbanization trends. Through the application of AI, urban infrastructure can be optimized by improving energy efficiency, streamlining transportation, and managing housing needs; AI makes it possible to reduce traffic congestion and advance mobility in transportation systems, such as prescriptive traffic management and autonomous vehicles (Regona et al., 2024).In cities like Singapore, AI manages real-time traffic and monitors energy consumption, setting urban efficiency benchmarks (Padhiary et al., 2025). On a similar note, Tennet TSO, a German transmission system operator, has been utilizing AI-based forecasting and IBM Watson's cognitive computing platform to anticipate renewable energy generation in real time, allowing real-time grid adjustments and maximizing clean energy use. (Mahadik, Sheetal et al., 2025) 3Nowadays,...
|
26248212
|
AI
|
10.3389/frai.2025.1426455
|
Analyzing handwriting legibility through hand kinematics
|
Introduction: Handwriting is a complex skill that requires coordination between human motor system, sensory perception, cognitive processing, memory retrieval, and linguistic proficiency. Various aspects of hand and stylus kinematics can affect the legibility of a handwritten text. Assessing handwriting legibility is challenging due to variations in experts' cultural and academic backgrounds, which introduce subjectivity biases in evaluations.Methods: In this paper, we utilize a deep-learning model to analyze kinematic features influencing the legibility of handwriting based on temporal convolutional networks (TCN). Fifty subjects are recruited to complete a 26-word paragraph handwriting task, designed to include all possible orthographic combinations of Arabic characters, during which the hand and stylus movements are recorded. A total of 117 different spatiotemporal features are recorded, and the data collected are used to train the model. Shapley values are used to determine the important hand and stylus kinematics features toward evaluating legibility. Three experts are recruited to label the produced text into different legibility scores. Statistical analysis of the top 6 features is conducted to investigate the differences between features associated with high and low legibility scores.Results: Although the model trained on stylus kinematics features demonstrates relatively high accuracy (around 76%), where the number of legibility classes can vary between 7 and 8 depending on the expert, the addition of hand kinematics features significantly increases the model accuracy by approximately 10%. Explainability analysis revealed that pressure variability, pen slant (altitude, azimuth), and hand speed components are the most prominent for evaluating legibility across the three experts.Discussion: The model learns meaningful stylus and hand kinematics features associated with the legibility of handwriting. The hand kinematics features are important for accurate assessment of handwriting legibility. The proposed approach can be used in handwriting learning tools for personalized handwriting skill acquisition as well as for pathology detection and rehabilitation.
|
26248212
|
AI
|
10.3389/feduc.2025.1562391
|
Digital learning in the 21st century: trends, challenges, and innovations in technology integration
|
The integration of digital technologies into education represents a significant evolution in the pedagogical landscape, with the potential to enhance accessibility, engagement, and personalization in learning. This review synthesizes current trends, challenges, and innovations within digital learning, emphasizing the impact of artificial intelligence (AI), virtual reality (VR), and online platforms on student achievement. It highlights the importance of addressing technical, pedagogical, and socioeconomic challenges to ensure equitable access to technology. Successful initiatives like the Open University illustrate digital learning's potential to improve educational outcomes. The review also anticipates future directions, including the expanding role of AI, VR, mobile learning, and blockchain in education. It concludes with strategic recommendations for educators and policymakers to adopt best practices, prioritize infrastructure development, and focus on continuous professional development to leverage the benefits of digital learning. As education enters an era of digital transformation, a collaborative approach among stakeholders will be essential in creating an inclusive and effective learning environment for the future.
|
2504284X
|
EDUCATION
|
10.3389/feduc.2025.1570389
|
Integrating artificial intelligence into pre-clinical medical education: challenges, opportunities, and recommendations
|
As AI technologies continue to advance and influence healthcare, it is imperative that medical education evolves to equip future physicians with the necessary skills and understanding of AI applications. In pre-clinical curricula, AI tools can streamline administrative processes, enhance teaching methodologies, and provide personalized learning experiences for students. Moreover, AI has the potential to shape students' professionalism, ethical decision-making, and critical thinking skills, which are essential for their future roles in the medical field. However, this shift also raises critical challenges, such as the need for ethical guidelines, adequate infrastructure, and avoiding overreliance on AI. Recent studies have shown a significant gap in formal AI education within medical studies, while the general attitudes of students towards AI is positive [1] This paper is based on a comprehensive narrative review of current literature and expert discussions. It aims to identify the key areas affected by AI integration in pre-clinical medical education, and to provide recommendations to maximize its benefits while mitigating associated risks.Universities face significant organizational challenges in integrating AI, such as ensuring ethical AI usage, addressing data privacy concerns, and meeting infrastructural requirements. Developing comprehensive policies to guide AI's role in academic settings is essential, including its application in exams, theses, teaching, and clinical decision-making tools [1][4]. Policy development requires good cooperation between academic self-governance bodies (academic freedom) and university management (management of personnel and financial resources, and enforcement of rules) [2].One significant opportunity presented by AI integration is the optimization of administrative tasks. AI-driven data analysis can improve resource management, allowing universities to allocate resources more efficiently and effectively [5][6]. Additionally, AI can enhance global collaboration in medical education by facilitating communication and information sharing across institutions worldwide. Implementing digital learning infrastructures, such as virtual simulation labs and AIassisted learning platforms, has the potential to improve teaching efficiency and provide innovative educational experiences for students [7][8].However, several challenges persist. Universities must invest in digital technologies while balancing traditional educational needs to prevent over-reliance on AI [8][9]. Ensuring equal and fair access to AI tools for all students is crucial to avoid disparities in educational opportunities. The freedom of teaching should not be impaired by university-wide adjustments. Establishing ethical frameworks is imperative to promote the responsible use of AI, especially concerning patient data in AI models [8][10] [11][12]. Addressing these ethical considerations is essential for safeguarding data privacy and upholding academic integrity.At the program management level, aligning the needs of students and educators regarding AI usage in education is crucial. This alignment impacts curriculum design, coordination among educators, and the preparation of students for future job roles in an AI-influenced healthcare environment [13].One of the primary challenges is ensuring that curricula remain relevant amid AI's growing impact on healthcare. This necessitates significant restructuring of courses to integrate AI tools while maintaining the essential human elements of medical education, such as patient interaction and ethical decision-making [10] [14][15] [16]. Program managers must balance the incorporation of new technologies with the preservation of core medical competencies to provide a comprehensive education under conditions in which these core competencies are continually reassessedConversely, AI offers opportunities to develop innovative course structures that include experiential components. For instance, AI simulations for clinical decision-making can enhance learning outcomes by providing students with practical, hands-on experience in a controlled environment. Additionally, AI can help identify future job market needs, enabling study programs to adapt their curricula and train students in AI-driven medical technologies [1][6][17]. This proactive approach ensures that graduates are better prepared for the evolving demands of the healthcare sector.For teaching staff, integrating AI into medical curricula demands the acquisition of new skills and methodologies, as well as an understanding of how to balance AI with traditional teaching methods. AI tools such as virtual tutors and simulation resources have the potential to greatly enhance the learning experience, but only when applied appropriately. Therefore, educators need to be trained adequately to use these tools effectively, and they need to be willing to receive that training [6][10].Opportunities for educators include utilizing AI-enhanced teaching tools, including virtual simulations and real-time feedback systems, which provide more personalized and adaptive learning experiences for students [14][19]. AI can also streamline assessment and grading processes, allowing educators to devote more time to developing students' higher-order thinking skills [5]. By reducing administrative burdens, educators can focus on facilitating critical analysis, problem-solving abilities, and clinical reasoning.Nevertheless, challenges persist. Training educators in AI usage is essential to ensure they can leverage these tools effectively and confidently integrate them into their teaching practices [5][8].Additionally, ethical concerns regarding the use of AI in education, such as issues of academic integrity and potential biases in AI algorithms, need to be carefully managed [18]. Educators must be equipped not only with technical skills but also with an understanding of the ethical implications of AI to guide students...
|
2504284X
|
EDUCATION
|
10.3390/cancers17071157
|
Estimating the Morbidity of Robot-Assisted Radical Cystectomy Using the Comprehensive Complication Index: Data from the Asian Robot-Assisted Radical Cystectomy Consortium
|
Background/Objectives: The Clavien–Dindo classification (CDC) grades the most severe post-operative complication and may not comprehensively reflect cumulative surgical morbidity. Our objective was to investigate the potential incremental role of the comprehensive complication index (CCI) over the CDC in defining the quality of robot-assisted radical cystectomy (RARC). Methods: Data were extracted from the Asian RARC Consortium database. Complications were classified using the CCI (CCI = 0, CCI < 75th and ≥75th percentile) and CDC. Adverse peri-operative outcomes such as length of stay >14 days (LOS > 14 days), estimated blood loss >350 mL (EBL > 350 mL), time to solid food intake >4 days (TFI > 4 days) and 30-day readmission rates were analyzed. The area under the curve (AUC) of the receiver operating characteristic (ROC) curves for CCI and CDC were compared for the various adverse outcomes. Results: The peri-operative complication rate was 44.4%, comprising 11.6% with severe complications (CDC ≥ III). The mean CCI was 10.2 (±13.5) while median CCI was 0 (IQR 0–21). There were 7.6% of patients with >one perioperative complication. On adjusted analysis, CCI ≥ 75th percentile was significantly associated with greater LOS (>14 days) (OR 2.21, 95% CI 1.47–3.31, p < 0.001) compared to when CCI = 0. There were no significant differences in the AUC between CDC and CCI in predicting LOS > 14 days, TFI > 4 days, 30-day readmission or EBL > 350 mL. Conclusions: In our multi-institutional cohort, the CCI did not provide additional discrimination over CDC, and this is likely related to the limited number of complications that occurred per individual in the Asian RARC cohort. Hence, the perceived advantages of CCI over CDC are contextual.
|
20726694
|
ONCOLOGY
|
10.3389/fonc.2025.1524714
|
Identification of MAD2L1 as a novel biomarker for hepatoblastoma through bioinformatics and machine learning approaches
|
Objective: This study aims to identify potential biomarkers for Hepatoblastoma (HB) using bioinformatics and machine learning, and to explore their underlying mechanisms of action.Methods: We analyzed the datasets GSE131329 and GSE133039 to perform differential gene expression analysis. Single-sample gene set enrichment analysis (ssGSEA) and weighted gene co-expression network analysis (WGCNA) were utilized to identify gene modules linked to gene set activity. Protein-protein interaction (PPI) networks were constructed to identify hub genes, while random forest and support vector machine models were employed to screen for key diagnostic genes. Survival and immune infiltration analyses were conducted to assess the prognostic significance of these genes. Additionally, the expression levels, biological functions, and mechanisms of action of the selected genes were validated in HB cells through relevant experimental assays.Results: We identified 1,377 and 1,216 differentially expressed genes in datasets GSE131329 and GSE133039, respectively. ssGSEA and WGCNA analyses identified 234 genes significantly linked to gene set activity. PPI analysis identified 20 core Hub genes. Machine learning highlighted three key diagnostic genes: CDK1, CCNA2, and MAD2L1. Studies have demonstrated that MAD2L1 is significantly overexpressed in HB and is associated with prognosis. WGCNA revealed that MAD2L1 is enriched in gene sets related to E2F_ TARGETS and G2M_CHECKPOINT. Experimental assays demonstrated that MAD2L1 knockdown significantly inhibits the proliferation, migration, and invasion of HB cell lines, and that MAD2L1 promotes cell cycle progression through the regulation of E2F.Conclusion: Our study identifies MAD2L1 as a novel potential biomarker for HB, providing new strategies for early diagnosis and targeted therapy in HB.
|
2234943X
|
ONCOLOGY
|
10.3390/cancers17071179
|
Inducing Targeted, Caspase-Independent Apoptosis with New Chimeric Proteins for Treatment of Solid Cancers
|
Background: Most newly developed anticancer treatments trigger tumor cell death through apoptosis, for which involvement of caspases activity is essential. However, numerous mutations in apoptotic pathways that lead to cancer and favor resistance to apoptosis are known; most are related to caspase-dependent apoptosis pathways and thus have low efficacy. To overcome these limitations, we constructed a novel chimeric protein, GnRH-AIF, using a gonadotropin-releasing hormone (GnRH) analog as a targeting moiety and the apoptosis-inducing factor (AIF) in its cleaved form as a killing moiety, fused at the cDNA level. AIF has a crucial role in the caspase-independent apoptotic pathway. A wide variety of solid tumors overexpress GnRH-receptors (GnRH-R) that are targeted by the new GnRH-AIF chimeric protein. Methods and Results: In this study, we constructed, expressed, and highly purified GnRH-AIF chimeric proteins. We demonstrated the ability of the chimera to enter and specifically and very efficiently kill solid cancer cell lines overexpressing GnRH-R. Most importantly, upon its entry, GnRH-AIFs translocate to the nucleus where it causes DNA fragmentation leading to a direct caspase-independent apoptotic death. As AIFs lack nuclease activity, our findings also emphasize that cell death induced by GnRH-AIF is dependent on the presence of the ENDOG and PPIA proteins, known to participate in the formation of a DNA–degradosome complex. Finally, we demonstrated the high anti-tumor efficacy of the GnRH-AIF ex vivo, in a human, colon cancer organoid model. Conclusions: Our study shows the potential of using a GnRH-AIF chimeric protein as a novel approach to treat solid cancers that overexpress GnRH-R, via a caspase-independent apoptotic pathway.
|
20726694
|
ONCOLOGY
|
10.3390/cancers17071191
|
Significance of 5-ALA-Guided Fluorescence in Resection of Invasive Intracranial Meningiomas: Findings from a Prospective Clinical Study
|
Background: In cases of intracranial meningiomas invading into surrounding tissues, determining the resection boundary can be challenging and often makes complete resection difficult. In such situations, the introduction of novel intraoperative techniques to identify infiltrative tumor components is desirable to improve the extent of tumor resection. Methods: A prospective clinical study was conducted on patients with intracranial meningiomas suspected of infiltration into the surrounding tissues. After completing the tumor resection under conventional white-light microscopy, intraoperative fluorescence diagnosis using 5-aminolevulinic acid (5-ALA) was performed to determine whether additional resection of the unintended residual tumor was feasible. Results: Intraoperative fluorescence diagnosis enabled additional resection of the residual tumor in 38.5% of the 13 enrolled cases and 45.5% of the 11 cases in which the tumor exhibited fluorescence positivity. Among the additional resected specimens, tumor infiltration was observed in all fluorescence-positive lesions of the bone and dura mater, whereas tumor cells were detected in only 33.3% of the fluorescence-positive areas in the adjacent brain parenchyma. Conclusions: Intraoperative fluorescence diagnosis using 5-ALA enhanced the extent of the resection of invasive meningiomas. Future large-scale studies are warranted to determine whether 5-ALA fluorescence diagnosis contributes to reducing tumor recurrence and improving overall survival in patients with invasive intracranial meningiomas.
|
20726694
|
ONCOLOGY
|
10.3389/feduc.2025.1550969
|
The impact of study abroad social interactions on post-return relationships with international students: Japanese students’ perceptions of recategorization
|
Previous research has focused on how students adapt to the host country during study abroad. However, less is known about how these experiences influence students’ social engagement upon returning home. This study explores how Japanese students’ social interactions abroad influence their relationships with international students in Japan after their return. Using a qualitative approach based on grounded theory, semi-structured interviews were conducted with 24 Japanese students who had studied abroad for one academic year. The findings suggest that social interactions abroad facilitate recategorization, a process in which individuals redefine group boundaries and develop a broader shared identity. This process was influenced by four key factors: language and social skills, motivation, opportunities, and perceived fit. Through this process, Japanese students expanded their group boundaries and formed a shared identity with international students in Japan as individuals with study abroad experience. As a result, they developed more positive attitudes toward international students, heightened empathy, and a stronger motivation to engage with and help international students. These findings indicate that recategorization can occur through the formation of a new social identity based on shared experiences rather than direct intergroup contact, highlighting the long-term impact of study abroad on students’ intercultural engagement. This study underscores Japanese students’ tendency to identify with international students in Japan rather than with host nationals upon their return.
|
2504284X
|
EDUCATION
|
10.3390/ejihpe15040050
|
Evaluating the Effects of Sensorimotor Training on the Physical Capacities of Older People
|
Background: Physical activity (PA) plays a crucial role in improving the quality of life (QoL) in older people, particularly by enhancing their balance and movement coordination. Objective: This study aimed to assess the effects of sensorimotor training intervention in older adults. Methods: A total of 90 participants, divided into a Control Group (n = 44) and Experimental Group (n = 46) were involved in a 24-week sensorimotor training program. The physical capacities of the participants were assessed both before and after the intervention program. Strength and flexibility were measured using the “Rikli and Jones” protocol (1999), while agility and speed were assessed through “Timed-up-and-go” tests. Taking into account the participants’ gender, a descriptive analysis of the sample was conducted to describe the data using the mean and standard deviation. Student’s T test was performed to compare the differences between the groups according to the first and second data collection moments (before and after the intervention). Jamovi software (v. 2.5.2.0) was used to develop the statistical analysis, using a p-value of less than 0.05 to assess the statistical significance. Results: The Experimental Group showed significant improvements across all the analyzed variables following the intervention (p < 0.005), indicating substantial gains in physical capacities. In contrast, the Control Group in the “sitting and reaching” test did not show a significant difference between the groups highlighting the lack of improvement without intervention. According to the effect size of the sample, it was observed that the parameters “reach behind your back (right)” and “reach behind your back (left)” showed the highest effect size comparing the Control Group and Experimental Group (ES: 0.60, 0.71). Conclusions: The findings highlight the practical clinical impact of implementing tailored physical activity programs for older adults. Such interventions are critical for enhancing QoL, reducing the risk of falls, injuries, and chronic illnesses, and promoting overall health, independence, and well-being. Integrating sensorimotor training into the routine care for older people can support healthy aging and functional independence.
|
22549625
|
PSYCHOLOGY
|
10.3389/frai.2025.1542320
|
An optimized system for predicting energy usage in smart grids using temporal fusion transformer and Aquila optimizer
|
This research presents an optimized system for predicting energy usage in smart grids by integrating the Temporal Fusion Transformer (TFT) with the Aquila Optimizer (AO). The study addresses the growing need for accurate energy consumption forecasts in smart grids, driven by the increasing adoption of renewable energy and real-time data collection through smart meters. The TFT model leverages self-attention mechanisms to handle complex time-series data, improving forecasting accuracy across various time horizons. To enhance predictive performance, the Aquila Optimizer, a nature-inspired algorithm, is employed to fine-tune critical hyperparameters, ensuring optimal model convergence and performance. The proposed AO-TFT model is evaluated against traditional models like LSTM and CNN-BiLSTM, demonstrating superior accuracy, lower RMSE, and faster computation times. The research also analyses the impact of various factors, including building types, weather conditions, and load variations on energy prediction. The proposed AO-TFT model achieved a significantly lower RMSE of 0.48 and MAE of 0.31, demonstrating superior accuracy compared to traditional models. Future work is suggested to explore hybrid optimization techniques and real-time adaptive models for dynamic grid management.
|
26248212
|
AI
|
10.3390/ai6040072
|
Multimodal Data Fusion for Tabular and Textual Data: Zero-Shot, Few-Shot, and Fine-Tuning of Generative Pre-Trained Transformer Models
|
In traffic safety analysis, previous research has often focused on tabular data or textual crash narratives in isolation, neglecting the potential benefits of a hybrid multimodal approach. This study introduces the Multimodal Data Fusion (MDF) framework, which fuses tabular data with textual narratives by leveraging advanced Large Language Models (LLMs), such as GPT-2, GPT-3.5, and GPT-4.5, using zero-shot (ZS), few-shot (FS), and fine-tuning (FT) learning strategies. We employed few-shot learning with GPT-4.5 to generate new labels for traffic crash analysis, such as driver fault, driver actions, and crash factors, alongside the existing label for severity. Our methodology was tested on crash data from the Missouri State Highway Patrol, demonstrating significant improvements in model performance. GPT-2 (fine-tuned) was used as the baseline model, against which more advanced models were evaluated. GPT-4.5 few-shot learning achieved 98.9% accuracy for crash severity prediction and 98.1% accuracy for driver fault classification. In crash factor extraction, GPT-4.5 few-shot achieved the highest Jaccard score (82.9%), surpassing GPT-3.5 and fine-tuned GPT-2 models. Similarly, in driver actions extraction, GPT-4.5 few-shot attained a Jaccard score of 73.1%, while fine-tuned GPT-2 closely followed with 72.2%, demonstrating that task-specific fine-tuning can achieve performance close to state-of-the-art models when adapted to domain-specific data. These findings highlight the superior performance of GPT-4.5 few-shot learning, particularly in classification and information extraction tasks, while also underscoring the effectiveness of fine-tuning on domain-specific datasets to bridge performance gaps with more advanced models. The MDF framework’s success demonstrates its potential for broader applications beyond traffic crash analysis, particularly in domains where labeled data are scarce and predictive modeling is essential.
|
26732688
|
AI
|
10.3390/ejihpe15040055
|
Ethical Climate, Intrinsic Motivation, and Affective Commitment: The Impact of Depersonalization
|
Although affective commitment has been the focus of numerous studies, we know relatively little about certain factors that drive or hinder its progress. In this sense, this study contributes to the knowledge on the subject by establishing a relationship between a benevolent ethical climate and affective commitment, taking into account the mediating effect of intrinsic motivation. Furthermore, we highlight depersonalization as an aspect that can hinder these relationships when it assumes a moderating function. The sample was established through 448 employees of the Colombian electrical sector. The mediating effect was confirmed through a four-step method. The moderated mediation model was examined using SEM structural equations. The results show that a benevolent ethical climate is positively related to affective commitment and that intrinsic motivation is a mediating factor that justifies this relationship. However, depersonalization moderates the mediation between benevolent ethical climate, intrinsic motivation, and affective commitment. Specifically, the positive effect of the benevolent ethical climate on affective commitment is halted when depersonalization is high. The positive relationship between intrinsic motivation and affective commitment is interrupted when depersonalization is medium or high. Finally, as depersonalization progresses, the positive relationship between a benevolent ethical climate and intrinsic motivation is reduced. Therefore, organizations in the Colombian electrical sector must take measures that, in addition to avoiding social isolation, behave as indicators that warn when employees’ behaviors change significantly.
|
22549625
|
PSYCHOLOGY
|
10.1186/s40359-020-00440-2
|
Evaluation of the effect of fatigue on the coping behavior of international truck drivers
|
Background: Fatigue can affect the behavior of drivers. While the driver must be able to respond and cope appropriately to the critical situations, which is known as the ability to cope with a crisis. It is likely that the fatigue can change the people’s coping style and thereby increase the chance of the crashes. Therefore, the present study aimed to investigate the effects of fatigue on the coping behavior of international truck drivers. Methods: This study was conducted on 239 of international truck drivers employed in Iran. The Endler and Parker coping strategies questionnaire (CISS) and Persian version of the Fatigue Multidimensional Fatigue Inventory (MFI) were used to evaluate the coping styles of the drivers and the drivers’ fatigue, respectively. Results: The mean values of the total fatigue before and after traveling were 36.77 and 76.13, respectively. The mean values of coping styles of the problem-oriented, emotion-oriented, and avoidance before traveling were 53.66, 40.91, and 38.17, respectively, and those after traveling were 45.59, 51.18, and 36.45, respectively. The chi-square test demonstrated that there was a significant difference in the coping style of drivers before and after the trip (P < 0.001), and the percent of individuals with emotion-oriented increased. Conclusions: In general, the results showed that fatigue due to traveling could change the coping styles of subjects from problem-oriented to emotion-oriented and avoidance. This can increase the statistics of driving accidents.
|
20507283
|
PSYCHOLOGY
|
10.3389/frai.2025.1543603
|
Emotional prompting amplifies disinformation generation in AI large language models
|
Introduction: The emergence of artificial intelligence (AI) large language models (LLMs), which can produce text that closely resembles human-written content, presents both opportunities and risks. While these developments offer significant opportunities for improving communication, such as in health-related crisis communication, they also pose substantial risks by facilitating the creation of convincing fake news and disinformation. The widespread dissemination of AI-generated disinformation adds complexity to the existing challenges of the ongoing infodemic, significantly affecting public health and the stability of democratic institutions.Rationale: Prompt engineering is a technique that involves the creation of specific queries given to LLMs. It has emerged as a strategy to guide LLMs in generating the desired outputs. Recent research shows that the output of LLMs depends on emotional framing within prompts, suggesting that incorporating emotional cues into prompts could influence their response behavior. In this study, we investigated how the politeness or impoliteness of prompts affects the frequency of disinformation generation by various LLMs.Results: We generated and evaluated a corpus of 19,800 social media posts on public health topics to assess the disinformation generation capabilities of OpenAI’s LLMs, including davinci-002, davinci-003, gpt-3.5-turbo, and gpt-4. Our findings revealed that all LLMs efficiently generated disinformation (davinci-002, 67%; davinci-003, 86%; gpt-3.5-turbo, 77%; and gpt-4, 99%). Introducing polite language to prompt requests yielded significantly higher success rates for disinformation (davinci-002, 79%; davinci-003, 90%; gpt-3.5-turbo, 94%; and gpt-4, 100%). Impolite prompting resulted in a significant decrease in disinformation production across all models (davinci-002, 59%; davinci-003, 44%; and gpt-3.5-turbo, 28%) and a slight reduction for gpt-4 (94%).Conclusion: Our study reveals that all tested LLMs effectively generate disinformation. Notably, emotional prompting had a significant impact on disinformation production rates, with models showing higher success rates when prompted with polite language compared to neutral or impolite requests. Our investigation highlights that LLMs can be exploited to create disinformation and emphasizes the critical need for ethics-by-design approaches in developing AI technologies. We maintain that identifying ways to mitigate the exploitation of LLMs through emotional prompting is crucial to prevent their misuse for purposes detrimental to public health and society.
|
26248212
|
AI
|
10.3389/frai.2025.1546064
|
Legal regulation of AI-assisted academic writing: challenges, frameworks, and pathways
|
Introduction: The widespread application of artificial intelligence in academic writing has triggered a series of pressing legal challenges.Methods: This study systematically examines critical issues, including copyright protection, academic integrity, and comparative research methods. We establishes a risk assessment matrix to quantitatively analyze various risks in AI-assisted academic writing from three dimensions: impact, probability, and mitigation cost, thereby identifying high-risk factors.Results: The findings reveal that AI-assisted writing challenges fundamental principles of traditional copyright law, with judicial practice tending to position AI as a creative tool while emphasizing human agency. Regarding academic integrity, new risks, such as “credibility illusion” and “implicit plagiarism,” have become prominent in AI-generated content, necessitating adaptive regulatory mechanisms. Research data protection and personal information security face dual challenges in data security that require technological and institutional innovations.Discussion: Based on these findings, we propose a three-dimensional regulatory framework of “transparency, accountability, technical support” and present systematic policy recommendations from institutional design, organizational structure, and international cooperation perspectives. The research results deepen understanding of legal attributes of AI creation, promote theoretical innovation in digital era copyright and academic ethics, and provide practical guidance for academic institutions in formulating AI usage policies.
|
26248212
|
AI
|
10.1186/s40359-025-02667-3
|
The role of mind wandering and anxiety in the association between internet addiction and hyperactivity-impulsivity: a serial mediation model
|
Hyperactivity-Impulsivity have significant negative effects on adolescents’ academic performance, physical and mental health, and social relationships. This study aims to deeply explore the relationship between Hyperactivity-Impulsivity in adolescents and Internet Addiction. Unlike previous studies, this study further explores a potential serial mediation model involving Mind Wandering and Anxiety. A total of 2042 adolescents completed assessments using the Internet Addiction Test (IAT), the Mind Wandering Questionnaire (MWQ), the Generalized Anxiety Disorder 2(GAD-2), and the ASRS short scale to evaluate Internet Addiction, Mind Wandering, Anxiety, and Hyperactivity-Impulsivity, respectively. Internet Addiction, Mind Wandering, and Anxiety significantly influence adolescents’ Hyperactivity-Impulsivity (p <.001). Mediation analysis further indicates that Internet Addiction is associated with Hyperactivity-Impulsivity through the serial mediating effects of Mind Wandering and Anxiety(p <.01). These findings highlight Mind Wandering and Anxiety as key mediators in the link between Internet Addiction and Hyperactivity-Impulsivity in adolescents. This study sheds light on how Internet Addiction influences Hyperactivity-Impulsivity among adolescents and underscores the importance of preventive measures. We recommend implementing interventions aimed at fostering healthy Internet usage habits and providing robust mental health support to safeguard adolescents’ physical and mental well-being.
|
20507283
|
PSYCHOLOGY
|
10.3389/fonc.2025.1560008
|
FGFR3-TACC3 fusion gene promotes glioblastoma malignant progression through the activation of STAT3 signaling pathway
|
Objective: The Fibroblast growth factor receptors 3-transforming acidic coiled-coil-containing protein 3 (FGFR3-TACC3, F3-T3) oncogenic fusion gene, identified in malignant tumors such as gliomas and bladder cancer, has been particularly noted in recurrent gliomas where it is considered to drive malignant progression, thus presenting itself as a viable therapeutic target. However, the precise mechanism by which F3-T3 facilitates the malignant progression of glioma is not fully understood.Methods: Correction analysis of STAT3 and FGFR3 with major glioma mutation types and pan-cancer analysis was conducted using The Cancer Genome Atlas (TCGA) database. A series of phenotypic experiments, including CCK-8, EdU, colony-formation assay, wound healing assay, and transwell assay were conducted to detect the effects of F3-T3 on proliferation, invasion, and migration of glioma cells. The association between F3-T3 and epithelial-mesenchymal transition (EMT) was investigated through enrichment analysis of the E-MTAB-6037 gene chip database and confirmed by western blot. The underling mechanism were further inferred and validated through RNA sequencing, E-MTAB-6037 gene chip data, and western blot. The relationship between p-STAT3 expression and the WHO grade of glioma was evaluated using immunohistochemistry (IHC) and tissue microarray analysis. Furthermore, the results of vivo experiments and IHC has confirmed the impact of F3-T3 on glioma malignant progression and activation of the STAT3 signaling pathway.Results: The experimental results from this study indicate that F3-T3 accelerates the epithelial-mesenchymal transition (EMT) process in glioma cells, thereby promoting their proliferation, invasion, and migration capabilities. Mechanistically, it was determined through RNA sequencing that the signal transducer and activator of transcription 3 (STAT3) signaling pathway is crucial for the malignant progression of F3-T3. This finding was further supported through follow-up experiments conducted after STAT3 knockdown. The role of the STAT3 pathway in gliomas was also reinforced through bioinformatic analysis and immunohistochemistry (IHC) on tissue microarrays (TMA). Further in vivo experiments corroborated the role of F3-T3 in enhancing glioma growth and progression.Conclusion: F3-T3 facilitates the proliferation, invasion, migration and EMT of glioma cells, thereby promoting their malignant progression through STAT3 signaling activation. These findings highlight its potential as a therapeutic target for glioma treatment.
|
2234943X
|
ONCOLOGY
|
10.3389/frai.2025.1478068
|
The role of AI for MRI-analysis in multiple sclerosis—A brief overview
|
Magnetic resonance imaging (MRI) has played a crucial role in the diagnosis, monitoring and treatment optimization of multiple sclerosis (MS). It is an essential component of current diagnostic criteria for its ability to non-invasively visualize both lesional and non-lesional pathology. Nevertheless, modern day usage of MRI in the clinic is limited by lengthy protocols, error-prone procedures for identifying disease markers (e.g., lesions), and the limited predictive value of existing imaging biomarkers for key disability outcomes. Recent advances in artificial intelligence (AI) have underscored the potential for AI to not only improve, but also transform how MRI is being used in MS. In this short review, we explore the role of AI in MS applications that span the entire life-cycle of an MRI image, from data collection, to lesion segmentation, detection, and volumetry, and finally to downstream clinical and scientific tasks. We conclude with a discussion on promising future directions.
|
26248212
|
AI
|
10.1186/s40359-025-02666-4
|
Effect of stressors on depressive mood among long-term high-altitude workers: a moderated mediation analysis
|
Diathesis-stress theory of depression is well known, which stresses that stressor is an inducing factor for depression in general population. High altitude, a combination of variety of stressors, is a special environment that may cultivate more depression. However, how different types of stressors contribute to depression and its underlying mechanisms in high-altitude populations remain unrevealed. The study aimed to reveal the effect of different stressors on depressive mood among long-term high-altitude workers in China and further explore the mediation of emotion regulation and moderation of parent-child alienation. 2065 Chinese workers at altitude of approximate 4200 m completed a cross-sectional survey with the Baker Depression Inventory-II scale, the Emotional Regulation scale, the Parent-child Alienation scale, and the Stressors scale (i.e., environmental factors, low social support, working challenges, accommodation, personal affairs, and cognitive factors). Correlation analysis showed positive correlations between stressors and depressive mood (r = 0.05–0.94, p < 0.05). Regression analysis indicated that low social support stressor was the strongest predictor of depressive mood (β = 0.21), while working challenges, personal affairs, and cognitive factors also positively predicted depressive mood. The mediating model showed that expression inhibition played a partial mediating (promoting) role between stressors and depressive mood, accounting for 3.13% of total variance. The moderating model showed that parent-child alienation played a moderating role in the model (β = 0.01, p < 0.001); a lower level of parent-child alienation effectively alleviated the impacts of stressors on depressive mood. Stressors (working challenges, personal affairs, cognitive factors, and especially low social support) positively predict the depressive mood of long-term high-altitude workers in China. Expression inhibition plays a promoting mediation in the relationship between stressors and depressive mood. A good parent-child relationship alleviates the negative impact of stressors on depressive mood. Findings provide new empirical support for diathesis-stress theory and attract further attention to less expression inhibition and better parent-child relationships in depression prevention.
|
20507283
|
PSYCHOLOGY
|
10.3390/ejihpe15040058
|
Lost in Thought or Just Lonely? Everyday Cognitive Competence in Middle Adulthood
|
Everyday cognitive competence refers to the ability to manage cognitively demanding tasks essential for maintaining functional independence. While cognitive abilities are well explored in explaining individual differences in everyday cognitive competence, growing attention has been directed toward the impact of non-cognitive factors like loneliness. This study aims to investigate how executive function (EF) components—updating, inhibition, and task shifting—predict everyday cognitive competence and whether loneliness explains the additional variance beyond EF processes. To account for the multifaceted nature of everyday cognitive competence, both performance-based (Everyday Problems Test—EPT) and self-reported measures (Cognitive Failures Questionnaire—CFQ) were administrated. The sample included 176 middle-aged adults (ages 43–65), a group suitable for investigating predictors of everyday cognitive competence in the early stages of cognitive aging. The findings reveal that updating is a significant predictor of the performance on the EPT, while loneliness is not. When self-reported cognitive lapses are considered, loneliness emerges as a significant predictor. The lack of a relationship between the EPT and CFQ, along with their differing associations with EF, loneliness, and sociodemographic factors, suggests they assess distinct aspects of everyday cognitive competence. This highlights the need for a multidimensional assessment framework to gain a comprehensive understanding of everyday cognitive competence in middle-aged adults.
|
22549625
|
PSYCHOLOGY
|
10.3390/ai6040074
|
Beautimeter: Harnessing GPT for Assessing Architectural and Urban Beauty Based on the 15 Properties of Living Structure
|
Beautimeter is a new tool powered by generative pre-trained transformer (GPT) technology, designed to evaluate architectural and urban beauty. Rooted in Christopher Alexander’s theory of centers, this work builds on the idea that all environments possess, to varying degrees, an innate sense of life. Alexander identified 15 fundamental properties, such as levels of scale and thick boundaries, that characterize living structure, which Beautimeter uses as a basis for its analysis. By integrating GPT’s advanced natural language processing capabilities, Beautimeter assesses the extent to which a structure embodies these 15 properties, enabling a nuanced evaluation of architectural and urban aesthetics. Using ChatGPT4o, the tool helps users generate insights into the perceived beauty and coherence of spaces. We conducted a series of case studies, evaluating images of architectural and urban environments, as well as carpets, paintings, and other artifacts. The results demonstrate Beautimeter’s effectiveness in analyzing aesthetic qualities across diverse contexts. Our findings suggest that by leveraging GPT technology, Beautimeter offers architects, urban planners, and designers a powerful tool to create spaces that resonate deeply with people. This paper also explores the implications of such technology for architecture and urban design, highlighting its potential to enhance both the design process and the assessment of built environments.
|
26732688
|
AI
|
10.3390/ai6040075
|
History-Aware Multimodal Instruction-Oriented Policies for Navigation Tasks
|
The rise of large-scale language models and multimodal transformers has enabled instruction-based policies, such as vision-and-language navigation. To leverage their general world knowledge, we propose multimodal annotations for action options and support selection from a dynamic, describable action space. Our framework employs a multimodal transformer that processes front-facing camera images, light detection and ranging (LIDAR) sensor’s point clouds, and tasks as textual instructions to produce a history-aware decision policy for mobile robot navigation. Our approach leverages a pretrained vision–language encoder and integrates it with a custom causal generative pretrained transformer (GPT) decoder to predict action sequences within a state–action history. We propose a trainable attention score mechanism to efficiently select the most suitable action from a variable set of possible options. Action options are text–image pairs and encoded using the same multimodal encoder employed for environment states. This approach of annotating and dynamically selecting actions is applicable to broader multidomain decision-making tasks. We compared two baseline models, ViLT (vision-and-language transformer) and FLAVA (foundational language and vision alignment), and found that FLAVA achieves superior performance within the constraints of 8 GB video memory usage in the training phase. Experiments were conducted in both simulated and real-world environments using our custom datasets for instructed task completion episodes, demonstrating strong prediction accuracy. These results highlight the potential of multimodal, dynamic action spaces for instruction-based robot navigation and beyond.
|
26732688
|
AI
|
10.3390/ai6040077
|
Enhancing the Classification of Imbalanced Arabic Medical Questions Using DeepSMOTE
|
The growing demand for telemedicine has highlighted the need for automated healthcare services, particularly in medical question classification. This study presents a deep learning model designed to address key challenges in telemedicine, including class imbalance and accurate routing of Arabic medical questions to the correct specialties. The model combines AraBERTv0.2-Twitter, fine-tuned for informal Arabic, with Bidirectional Long Short-Term Memory (BiLSTM) networks to capture deep semantic relationships in medical text. We used a labeled dataset of 5000 Arabic consultation records from Altibbi, covering five key medical specialties selected for their clinical relevance and frequency. The data underwent preprocessing to remove noise and normalize text. We employed stratified sampling to ensure representative distribution across the selected medical specialties. We evaluate multiple models using macro precision, macro recall, macro F1-score, weighted F1-score, and G-Mean. Our results demonstrate that DeepSMOTE combined with cross-entropy loss achieves the best performance. The findings offer statistically significant improvements and have practical implications for improving screening and patient routing in telemedicine platforms.
|
26732688
|
AI
|
10.1186/s40359-025-02682-4
|
Addressing the associative stigma of psychiatry and psychiatrists: a survey on the attitudes of medical and nursing students and doctors in Verona, Italy
|
Negative societal attitudes toward mental health often contribute to misconceptions and stereotypes about psychiatry, a phenomenon known as “associative stigma”. This stigma can hinder collaboration between psychiatrists and other specialists and deter students from pursuing psychiatry as a career. This study focused on one of the three main components of stigma by examining attitudes toward psychiatry and psychiatrists among medical and nursing students, as well as doctors, and identifying factors that influence these attitudes. A cross-sectional survey was conducted among medical and nursing students at the University of Verona and doctors affiliated to the Medical Professional Association of Verona. Attitudes toward psychiatry were assessed using the Attitude to Psychiatry Scale. Regression analysis evaluated the relationship between participants’ characteristics and their attitudes toward psychiatry and psychiatrists. A total of 511 medical students, 394 nursing students, and 638 doctors participated in the study. While students had generally positive attitudes towards psychiatry, they perceived it as lacking full respect within medial community (84% medical, 76% nursing), having low prestige (63.5% medical, 65.9% nursing), and receiving insufficient encouragement in university courses (39% medical, 41.7% nursing). Doctors also expressed positive attitudes, though to a lesser extent than students. Their primary concerns related to patient care: 81% reported feeling emotionally drained when treating psychiatric patients, and 58.2% felt that patients were not appreciative of the care received. Female students and doctors, students who had taken psychiatric courses, and doctors in non-surgical specialties exhibited more positive attitudes. This study revealed generally positive attitudes towards psychiatry, underscoring its relevance as a medical specialty. However, concerns regarding the discipline’s perceived status and respect within the medical field highlight areas for targeted interventions to enhance its image and encourage greater interest among students and professionals.
|
20507283
|
PSYCHOLOGY
|
10.1007/s00432-025-06176-z
|
Revealing tumor microenvironment communication through m6A single-cell analysis and elucidating immunotherapeutic potentials for cutaneous melanoma (CM)
|
Background The methylation of N6-methyladenosine (m6A) RNA plays a crucial role in the genetic regulation of various cancers. While m6A modifications have been extensively studied in the tumor microenvironment (TME) of several malignancies, their role in cutaneous melanoma (CM) remains unexplored. Methods Using Non-negative matrix factorization (NMF) analysis on single-cell RNA-seq data (GSE215121) from three CM samples obtained from public databases, 26 m6A RNA methylation regulators were utilized to determine TME subclusters, their expression, and function. Results Six distinct TME cell types were identified and NMF clustering further revealed unique m6A-based subpopulations of cancer-associated fibroblasts and T cells. The prognostic model demonstrated strong predictive capabilities, particularly for fibroblast and T cell m6A clusters, and highlighted COL3A1 as a critical regulator of melanoma-fibroblast interactions. Conclusion Highlighting the COL3A1 gene as a critical link and potential therapeutic target in melanoma could offer new avenues for targeted therapies and improve prognostic assessments in cutaneous melanoma.
|
14321335
|
ONCOLOGY
|
10.3390/cancers17081299
|
“Somewhat of an Adult”: Understanding the “Dance” of Competing Tensions Parents Manage While Caring for an Adolescent or Young Adult (AYA) Diagnosed with Hematologic Malignancy
|
Background: Parents supporting AYAs with blood cancer juggle dual, competing roles as cancer caregiver and parent, which may heighten distress as they feel pulled simultaneously in two opposing directions. Likewise, AYAs encounter paradoxical needs as they revert to being more dependent on their parents to prioritize their survival while their developmental trajectory toward independence is disrupted. Parents need help understanding the underlying tensions they face in caregiving to reduce their distress and promote their connectedness with their AYA. Using a dialectical lens, we identified tensions parents encountered while caregiving in three contexts (clinical, family, and online communication) to inform a targeted psychosocial intervention. Methods: In partnership with The Leukemia & Lymphoma Society, we recruited 20 parents for in-depth interviews. Parents cared for adolescents aged 15–18 (n = 10) or emerging adults aged 19–29 (n = 10) diagnosed >3 months prior and in active treatment or within 2 years since treatment ended. Transcripts were thematically analyzed. Results: Parents described four ongoing tensions they needed to negotiate as they cared for their AYA: (1) being the driver versus passenger in their child’s care; (2) coping with cancer together as a family versus separately; (3) deciding to reveal versus conceal information; and (4) expecting normative developmental and disease trajectories versus disrupted trajectories. These tensions characterize the complex caregiving “dance” parents navigate in all three care contexts. Conclusions: Psychosocial education can normalize these tensions for parents to promote healthier coping and reduce distress while enhancing connectedness with their AYA. As caregiver–patient outcomes are interrelated, it may improve AYAs’ well-being.
|
20726694
|
ONCOLOGY
|
10.3390/cancers17081333
|
Oncologic Outcomes of Young Breast Cancer Patients According to Tumor Biology
|
Background/Objectives: Young women frequently present with more aggressive breast cancer tumors. This retrospective study analyzed the oncological outcomes of patients under the age of 40 according to the tumor biology. Methods: Group comparisons were performed via the log-rank test. Recurrence and survival rates are presented according to the Kaplan–Meier method. Results: In total, 88 women (mean age 36) were included, but two presented with bilateral cancer, resulting in 90 tumors. Triple-negative carcinoma was most common, with 26.7% (n = 24); 11.1% (n = 10) were luminal A; 23.3% (n = 21) were luminal B HER2-negative; 15.6% (n = 14) were luminal B HER2-positive; and 6.7% (n = 6) were HER2-positive (non-luminal). Moreover, 26.1% (n = 23) of patients experienced recurrence (mean 40 months), with the highest recurrence rate in the HER2-positive (50%) and triple-negative (30.4%) groups. The 3- and 5-year recurrence-free survival rates were 84.9% and 77.3%, and the overall survival rates were 93.1% and 90.3%, respectively. No statistically significant differences in oncological outcomes were observed (p = 0.164). Conclusions: The results show that young women tend to have triple-negative and fast-growing breast carcinomas, with worse overall survival in the triple-negative group. More research is needed on the pathomechanisms of breast cancer development in young women, especially those leading to disease progression and resistance to therapy.
|
20726694
|
ONCOLOGY
|
10.3390/ejihpe15040063
|
Psychometric Properties of the Adverse Childhood Experiences Abuse Short Form (ACE-ASF) for Ecuadorian Youth
|
Adverse childhood experiences, such as abuse, are a risk factor for mental health and poor socio-emotional development in adulthood. Assessing these experiences in specific populations allows for the identification of patterns and the implementation of preventive interventions. Objective: To evaluate the psychometric properties of the abbreviated version of the Adverse Childhood Experiences Abuse Form (ACE-ASF) in Ecuadorian youth, aiming to ensure the validity, reliability, and consistency of the instrument in accurately measuring abuse dimensions in this Ecuadorian population. Methodology: An instrumental study was conducted on the psychometric properties of the eight-item ACE-ASF, applying it to a sample of 840 university students (52.1% females and 47.9% males). The evaluation focused on analyzing the factorial structure and internal consistency of the instrument in this sample. Results: The two-factor model showed a satisfactory fit across all levels of invariance (configural, metric, scalar, and strict), with acceptable fit indices (CFI, TLI, GFI, RMSEA, and SRMR). The internal consistency was adequate, as assessed using the McDonald’s omega and Cronbach’s alpha coefficients. Convergent and discriminant validity were confirmed using the AVE and HTMT indices, ensuring proper differentiation between the dimensions assessed. Conclusion: The ACE-ASF proved to be a valid and reliable instrument for assessing abuse experiences in Ecuadorian youth. Its two-factor structure reflects distinct yet related dimensions, providing a useful tool for identifying adverse childhood experiences in this population.
|
22549625
|
PSYCHOLOGY
|
10.3390/ejihpe15040064
|
Teachers’ Perceptions and Preparedness for Teaching English as a Foreign Language to Students with Developmental Dyslexia: A Systematic Review
|
Students with developmental dyslexia (DD) face significant challenges when learning English as a foreign language (EFL), highlighting the need for targeted support in educational systems. EFL teachers’ perceptions and preparedness regarding DD are crucial for effective instruction and improved learning outcomes in inclusive classrooms. However, no systematic review has yet explored EFL teachers’ perceptions and preparedness to teach students with DD. This systematic review, conducted in accordance with the PRISMA guidelines, examines existing research between 2005 and 2025 on EFL teachers’ perceptions and preparedness to teach students with DD. Studies were retrieved from databases including APA PsycNet, Crossref, ERIC, ProQuest, PubMed, and Scopus databases. Of 17,798 results, 16 studies met the inclusion criteria. The findings reveal mixed EFL teachers’ perceptions toward DD and inadequate training specific to DD. Moreover, practical teaching strategies and targeted interventions remain underrepresented in the literature. Most teachers lack formal DD-specific training, leading to insufficient classroom support. This review emphasizes the urgent need for improved in-service training and the development of effective resources. Future research should prioritize developing and evaluating practical teaching strategies and professional development programs on teacher preparedness in EFL contexts.
|
22549625
|
PSYCHOLOGY
|
10.1186/s40359-025-02548-9
|
A psycho-behavioral perspective research for elderly leisure sports participation via big-data and comparative analyses
|
The health of the elderly and the need for research to support them has never been more important. This study aims (a) to analyze the participation behavior of the elderly in leisure sports through big-data analysis and (b) to compare and analyze the motivations, limitations, and satisfaction of participation in leisure sports by age group. First, big-data analysis using text-mining technique was conducted using the TEXTOM program to collect and analyze data between May 1, 2023 and November 24, 2024. Next, a survey was conducted among adults aged 20 years and older who regularly participate in leisure sports to determine their motivations, limitations, and satisfaction with leisure participation. From June to December 2024, the data of 301 participants were collected and analyzed using SPSS 28.0. Specifically, this study analyzed the validity and reliability of the data and then compared and analyzed the three age groups through multivariate analysis of variance. Big-data analysis identified key terms and four clusters related to senior leisure sports participation: (a) Policy, (b) Welfare, (c) Senior Sports, and (d) Employment. The results of the comparative study through the questionnaire showed that compared to younger participants in leisure sports, the elderly showed higher results in the factors of self-challenge motive, social interaction motive, and leisure participation satisfaction, but lower results in the factor of cost constraints. This means that the elderly participate in leisure sports for challenge and social interaction, are more satisfied, and are less constrained by cost. The scientific and objective results of this study could be used as a resource to specifically understand the leisure sports participation behavior of the elderly.
|
20507283
|
PSYCHOLOGY
|
10.1186/s40359-025-02492-8
|
Empowering leadership and occupational burnout: the moderated mediation model
|
This study examines the impact of empowering leadership on occupational burnout through the mediating role of workaholism and the moderating effect of psychological hardiness in the relationship between empowering leadership and occupational burnout. The present study employs empowerment and hardiness theory. Further, the moderated mediation hypothesis was also investigated. Survey responses from 212 permanent employees (nurses) in the healthcare industry were gathered using the temporal separation (two time-lags with one month between the first and second lags) to test the proposed hypotheses. Different statistical analysis techniques, confirmatory factor analysis, discriminant and convergent validity and PROCES-macro were used. The current study findings shows that empowering leadership significantly reduces occupational burnout. Furthermore, the results of the study confirm that workaholism plays a crucial role as a mediator between empowering leadership and occupational burnout in the workplace. Additionally, the findings shows that empowering leadership burdens nurses by making them work excessively, which causes occupational burnout in the workplace. Furthermore, psychological hardiness is a significant moderator in the relationship between workaholism and occupational burnout. Finally, the moderated mediation model results showed that nurses with high psychological hardiness adjust and manage well with intense workloads, i.e., workaholism, when emboldened through their leaders which leads to reduction in occupational burnout. The findings emphasize the potential advantages and hazards of empowering leadership in the nursing profession and the management of healthcare. This study builds on earlier research by empirically investigating how workaholism and psychological hardiness influence the relationship between empowering leadership and occupational burnout in the nursing profession of Pakistan.
|
20507283
|
PSYCHOLOGY
|
10.3389/fonc.2025.1513774
|
Efficacy and safety of immune checkpoint inhibitors for brain metastases of non-small cell lung cancer: a systematic review and network meta-analysis
|
Background: Previous studies have demonstrated that immune checkpoint inhibitors (ICIs) significantly improve prognosis in lung cancer patients with brain metastases (BMs). This systematic review and network meta-analysis aims to evaluate the efficacy and safety of 10 ICIs recommended by the 2024 Chinese Society of Clinical Oncology guidelines for treating non-small cell lung cancer (NSCLC) without driver genes, focusing on NSCLC patients presenting with BMs.Materials and methods: A comprehensive literature search of PubMed, Embase, and the Cochrane Library was conducted through June 2024 to identify eligible controlled trials and head-to-head randomized controlled trials investigating 10 ICIs in NSCLC patients with BMs. Pairwise and network meta-analyses were performed using hazard ratios (HRs) and relative risks (RRs) with 95% confidence intervals (CIs). Treatment efficacy was ranked hierarchically through the surface under the cumulative ranking curve (SUCRA).Results: Sixteen trials from 11 studies, encompassing 1,274 NSCLC patients with BMs, were included. The meta-analysis demonstrated that ICIs significantly improved overall survival (OS: HR, 0.66; 95% CI, 0.52–0.85; P = 0.001) and progression-free survival (PFS: HR, 0.67; 95% CI, 0.54–0.84; P < 0.001). SUCRA ranking identified pembrolizumab as the most effective agent for OS improvement (SUCRA 71%), while camrelizumab showed superior PFS benefits (SUCRA 92%). ICIs were associated with increased objective response rates (RR: 1.52; 95% CI, 1.13–2.06; P = 0.006), but elevated risks of immune-mediated adverse events (RR: 2.50; 95% CI, 1.46–4.30; P = 0.001) and grade 3–5 immune-mediated adverse events and infusion reaction (RR: 6.39; 95% CI, 1.53–26.69; P = 0.011).Conclusion: ICIs demonstrate superior survival benefits compared to chemotherapy in NSCLC patients with BMs, with pembrolizumab and camrelizumab emerging as optimal choices for OS and PFS improvement, respectively. However, vigilant monitoring of immune-mediated adverse events and infusion reactions remains critical in clinical practice.
|
2234943X
|
ONCOLOGY
|
10.3390/ai6040081
|
The Impact of Ancient Greek Prompts on Artificial Intelligence Image Generation: A New Educational Paradigm
|
Background/Objectives: This article explores the use of Ancient Greek as a prompt language in DALL·E 3, an Artificial Intelligence software for image generation. The research investigates three dimensions of Artificial Intelligence’s ability: (a) the sense and visualization of the concept of distance, (b) the mixing of representational as well as mythic contents, and (c) the visualization of emotions. More specifically, the research not only investigates AI’s potentialities in processing and representing Ancient Greek texts but also attempts to assess its interpretative boundaries. The key question is whether AI can faithfully represent the underlying conceptual and narrative structures of ancient literature or whether its representations are superficial and constrained by algorithmic procedures. Methods: This is a mixed-methods experimental research design examining whether a specified Artificial Intelligence software can sense, understand, and graphically represent linguistic and conceptual structures in the Ancient Greek language. Results: The study highlights Artificial Intelligence’s possibility in classical language education as well as digital humanities regarding linguistic complexity versus AI’s power in interpretation. More specifically, the research not only investigates AI’s potentialities in processing and representing Ancient Greek texts but also attempts to assess its interpretative boundaries. The key question is whether AI can faithfully represent the underlying conceptual and narrative structures of ancient literature or whether its representations are superficial and constrained by algorithmic procedures. The study highlights Artificial Intelligence’s possibility in classical language education as well as digital humanities regarding linguistic complexity versus AI’s power in interpretation. Conclusions: The research is a step toward a more extensive discussion on Artificial Intelligence in historical linguistics, digital pedagogy, as well as aesthetic representation by Artificial Intelligence environments.
|
26732688
|
AI
|
10.1186/s40359-025-02707-y
|
Navigating stigma and somatization: a qualitative exploration of mental health experiences among middle-aged adults in rural China
|
This study investigated the experiences of stigma and somatization among middle-aged adults with mental health issues. Using frameworks of public stigma, self-stigma, affiliate stigma, and somatization (both presenting and functional), the study explores how individuals navigate the stigma associated with mental health. Interviews were conducted with middle-aged adults in rural areas, and the data were analyzed using Interpretative Phenomenological Analysis (IPA) to gain insights into their lived experiences. The findings reveal that mental health stigma in rural China significantly influences how individuals express mental distress, often leading to somatization. Patients tend to frame their mental health issues in terms of physical symptoms, such as headaches or fatigue, to avoid stigma. The study also highlights the role of cultural norms in shaping these expressions, particularly within the context of close-knit rural communities where mental health issues is stigmatized. The implications for education and policy are discussed, emphasizing the need for improved public mental health education and more equitable distribution of healthcare resources between urban and rural areas. This study contributes to the understanding of mental health stigma in rural China and offers practical suggestions for addressing mental health challenges in underserved communities.
|
20507283
|
PSYCHOLOGY
|
10.3390/educsci15040510
|
Development and Validation of a Questionnaire on Students’ Mathematics Capital: A Tool to Explore Opportunities in the Mathematics Classroom
|
Understanding students’ opportunities in mathematics education requires tools that capture the social and cultural dimensions shaping their engagement with the subject. One way to conceptualise these opportunities is through the notion of mathematics capital, which encompasses the resources and dispositions that students bring to their mathematical experiences. This study introduces and validates a questionnaire designed to measure secondary students’ mathematics capital, adapting the well-established science capital framework to the mathematical domain. Grounded in Bourdieu’s concept of capital, the questionnaire operationalises mathematics capital across mathematical forms of cultural capital, mathematics-related behaviours and practices, and mathematics-related forms of social capital. The questionnaire was administered to 119 students in an Italian secondary school as part of a broader study on mathematical memes. Statistical analyses, including correlation tests and Cronbach’s alpha, confirm the instrument’s reliability and internal coherence, highlighting the influence of both school and extracurricular environments. The questionnaire provides educators with a practical tool to better understand students’ engagement with mathematics and to inform strategies for fostering equity in mathematics education. By making mathematics capital a measurable construct, this research contributes to discussions on how cultural and social factors shape students’ trajectories in mathematics and beyond.
|
22277102
|
EDUCATION
|
10.3390/ai6040084
|
Artificial Intelligence in Ovarian Cancer: A Systematic Review and Meta-Analysis of Predictive AI Models in Genomics, Radiomics, and Immunotherapy
|
Background/Objectives: Artificial intelligence (AI) is increasingly influencing oncological research by enabling precision medicine in ovarian cancer through enhanced prediction of therapy response and patient stratification. This systematic review and meta-analysis was conducted to assess the performance of AI-driven models across three key domains: genomics and molecular profiling, radiomics-based imaging analysis, and prediction of immunotherapy response. Methods: Relevant studies were identified through a systematic search across multiple databases (2020–2025), adhering to PRISMA guidelines. Results: Thirteen studies met the inclusion criteria, involving over 10,000 ovarian cancer patients and encompassing diverse AI models such as machine learning classifiers and deep learning architectures. Pooled AUCs indicated strong predictive performance for genomics-based (0.78), radiomics-based (0.88), and immunotherapy-based (0.77) models. Notably, radiogenomics-based AI integrating imaging and molecular data yielded the highest accuracy (AUC = 0.975), highlighting the potential of multi-modal approaches. Heterogeneity and risk of bias were assessed, and evidence certainty was graded. Conclusions: Overall, AI demonstrated promise in predicting therapeutic outcomes in ovarian cancer, with radiomics and integrated radiogenomics emerging as leading strategies. Future efforts should prioritize explainability, prospective multi-center validation, and integration of immune and spatial transcriptomic data to support clinical implementation and individualized treatment strategies. Unlike earlier reviews, this study synthesizes a broader range of AI applications in ovarian cancer and provides pooled performance metrics across diverse models. It examines the methodological soundness of the selected studies and highlights current gaps and opportunities for clinical translation, offering a comprehensive and forward-looking perspective in the field.
|
26732688
|
AI
|
10.3390/cancers17081356
|
Intraoperative Radiation Therapy (IORT) in Gynecologic Cancers: A Scoping Review
|
Objective: We aimed to analyze the current literature for IORT in gynecological cancers and summarized clinical outcomes regarding patient selection. Methods: A systematic search was conducted utilizing PUBMED, Embase, and CINAHL to identify studies following PRISMA-ScR guidelines. A PICOS structure was utilized: population: patients with epithelial gynecological cancers; intervention: IORT; C: a comparator was not required, as we aimed to analyze patient selection; outcome: clinical outcomes and overall survival; and S: experimental and quasi-experimental analytical observational studies and descriptive observational studies, excluding case series published in English and limited to the last 10 years. Data extraction was conducted for patient selection, IORT, oncological outcomes, and morbidity. Results: A total of 707 results were identified, and 509 studies were uploaded to Covidence for screening after removing duplications. Of the 21 eligible studies, 9 were included in the final review. The total number of patients included was 348. The studies were retrospective single-institution studies, except for one. There was significant heterogeneity in their design and protocols. IORT was exclusively used for recurrent and advanced stage gynecological cancers adjunct to pelvic exenteration or laterally extended endopelvic resections with variable indications across institutions. The mean number of IORT patients per study was 2.8 per year. Survival rates were variable and dependent on the surgical margin. Endometrial cancer had a favorable outcome compared to vulvar and cervical cancers. Conclusions: Current clinical practice, as demonstrated by the research, is consistent with NCCN guidelines that endorse the application of IORT in instances of recurrent cervical, vaginal, and vulvar malignancies; however, there are no established recommendations for primary tumors. The analysis shows that there are gaps in our knowledge, mainly regarding the status of the margins, the criteria used to choose patients, and the outcomes that are specific to each histology. The standardization of protocols and prospectively powered studies are needed to refine patient selection criteria.
|
20726694
|
ONCOLOGY
|
10.3390/ai6040085
|
CacheFormer: High-Attention-Based Segment Caching
|
Efficiently handling long contexts in transformer-based language models with low perplexity is an active area of research. Numerous recent approaches like Linformer, Longformer, Performer, and Structured state space models (SSMs), have not fully resolved this problem. All these models strive to reduce the quadratic time complexity of the attention mechanism while minimizing the loss in quality due to the effective compression of the long context. Inspired by the cache and virtual memory principle in computers, where in case of a cache miss, not only the needed data are retrieved from the memory, but the adjacent data are also obtained, we apply this concept to handling long contexts by dividing it into small segments. In our design, we retrieve the nearby segments in an uncompressed form when high segment-level attention occurs at the compressed level. Our enhancements for handling long context include aggregating four attention mechanisms consisting of short sliding window attention, long compressed segmented attention, dynamically retrieving top-k high-attention uncompressed segments, and overlapping segments in long segment attention to avoid segment fragmentation. These enhancements result in an architecture that outperforms existing SOTA architectures with an average perplexity improvement of 8.5% over similar model sizes.
|
26732688
|
AI
|
10.3390/educsci15040511
|
ChatGPT in Education: Challenges in Local Knowledge Representation of Romanian History and Geography
|
The integration of AI tools like ChatGPT in education has sparked debates on their benefits and limitations, particularly in subjects requiring region-specific knowledge. This study examines ChatGPT’s ability to generate accurate and contextually rich responses to assignments in Romanian history and geography, focusing on topics with limited digital representation. Using a document-based analysis, this study compared ChatGPT’s responses to local archival sources, monographs, and topographical maps, assessing coverage, accuracy, and local nuances. Findings indicate significant factual inaccuracies, including misidentified Dacian tribes, incorrect historical sources, and geographic errors such as misplaced landmarks, elevation discrepancies, and incorrect infrastructure details. ChatGPT’s reliance on widely digitized sources led to omissions of localized details, highlighting a fundamental limitation when applied to non-digitized historical and geographic topics. These results suggest that while ChatGPT can be a useful supplementary tool, its outputs require careful verification by educators to prevent misinformation. Future research should explore strategies to improve AI-generated educational content, including better integration of regional archives and AI literacy training for students and teachers. The study underscores the need for hybrid AI-human approaches in education, ensuring that AI-generated text complements rather than replaces verified academic sources.
|
22277102
|
EDUCATION
|
10.1186/s40359-025-02729-6
|
Spiritual orientation and mental health: an SEM analysis of meaning and death attitudes as mediators in Turkish religious officials
|
This study examines the relationships between spiritual orientation, meaning in life, attitudes towards death, and indicators of psychological health (depression, anxiety, and stress) among 348 Muslim religious officials in Turkey (28% female). Using structural equation modelling (SEM), the results showed that spiritual orientation directly and indirectly reduces psychological distress by enhancing personal meaning and fostering more accepting attitudes towards death. Results showed a moderate positive association between spiritual orientation and meaning in life, and weak but significant negative associations between meaning/attitudes towards death and psychological symptoms. As one of the first empirical studies to examine the mediating role of death attitudes in this population, the research highlights the theoretical relevance of existential frameworks such as logotherapy. The study offers practical implications for the development of culturally sensitive psychoeducational and spiritual counselling programmes aimed at supporting the mental health of religious professionals exposed to grief and death-related stressors.
|
20507283
|
PSYCHOLOGY
|
10.1186/s40594-025-00543-5
|
Study of an effective machine learning-integrated science curriculum for high school youth in an informal learning setting
|
This study evaluates the effectiveness of a machine learning (ML) integrated science curriculum implemented within the Science Research Mentorship Program (SRMP) for high school youth at the American Museum of Natural History (AMNH) over 2 years. The 4-week curriculum focused on ML knowledge gain, skill development, and self-efficacy, particularly for under-represented youth in STEM. ML is increasingly prevalent in STEM fields, making early exposure to ML methods and artificial intelligence (AI) literacy crucial for youth pursuing STEM careers. However, STEM fields, particularly those focused on AI research and development, suffer from a lack of diversity. Learning experiences that support the participation of under-represented groups in STEM and ML are essential to addressing this gap. Participant learning was assessed through pre- and post-surveys measuring ML knowledge, skills, and self-efficacy. Results from the implementation of the curriculum show that participants gained understanding of ML knowledge and skills (p < 0.001, d = 1.083) and self-efficacy in learning ML concepts (p = 0.004, d = 0.676). On average, participants who identified as female and non-white showed greater learning gains than their white male peers (ML knowledge: p < 0.001, d = 1.191; self-efficacy: p = 0.006, d = 0.631), decreasing gaps in ML knowledge, skills, and self-efficacy identified in pre-survey scores. The ML-integrated curriculum effectively enhances students’ understanding and confidence in ML concepts, especially for under-represented groups in STEM, and provides a model for future ML education initiatives in informal science settings. We suggest that policy makers and school leaders take into account that high school age youth can learn ML concepts through integrated curricula while maintaining an awareness that curriculum effectiveness varies across demographic groups.
|
21967822
|
EDUCATION
|
10.3389/frai.2025.1553220
|
Handling missing data of using the XGBoost-based multiple imputation by chained equations regression method
|
This study introduces an XGBoost-MICE (Multiple Imputation by Chained Equations) method for addressing missing data in mine ventilation parameters. Using historical ventilation system data from Shangwan Coal Mine, scenarios with different missing rates (5, 10, and 15%) and iteration numbers (30 and 50) were simulated to validate the accuracy and effectiveness of the approach. The results demonstrate that as the missing rate increased from 5 to 15%, the Mean Squared Error (MSE) rose from 0.0445 to 0.3254, while the Explained Variance decreased from 0.988309 to 0.943267. Additionally, the Mean Absolute Error (MAE) increased by 0.29. Iteration experiments on the “frictional resistance per 100 meters” attribute showed convergence of MSE and MAE after six iterations. Overall, the XGBoost-MICE method exhibited high imputation accuracy and stable convergence across various missing data scenarios, providing robust technical support for optimizing intelligent mine ventilation systems.
|
26248212
|
AI
|
10.3389/frai.2025.1458707
|
Detection and classification of ChatGPT-generated content using deep transformer models
|
Introduction: The rapid advancement of AI, particularly artificial neural networks, has led to revolutionary breakthroughs and applications, such as text-generating tools and chatbots. However, this potent technology also introduces potential misuse and societal implications, including privacy violations, misinformation, and challenges to integrity and originality in academia. Several studies have attempted to distinguish and classify AI-generated textual content from human-authored work, but their performance remains questionable, particularly for AI models utilizing large language models like ChatGPT.Methods: To address this issue, we compiled a dataset consisting of both human-written and AI-generated (ChatGPT) content. This dataset was then used to train and evaluate a range of machine learning and deep learning models under various training conditions. We assessed the efficacy of different models in detecting and classifying AI-generated content, with a particular focus on transformer-based architectures.Results: Experimental results demonstrate that the proposed RoBERTa-based custom deep learning model achieved an F1-score of 0.992 and an accuracy of 0.991, followed by DistilBERT, which yielded an F1-score of 0.988 and an accuracy of 0.988. These results indicate exceptional performance in detecting and classifying AI-generated content.Discussion: Our findings establish a robust baseline for the detection and classification of AI-generated textual content. This work marks a significant step toward mitigating the potential misuse of AI-powered text generation tools by providing a reliable approach for distinguishing between human and AI-generated text. Future research could explore the generalizability of these models across different AI-generated content sources and address evolving challenges in AI text detection.
|
26248212
|
AI
|
10.3389/fonc.2025.1522237
|
Family resilience in patients with gynecological malignant tumors after radical hysterectomy: based on the Walsh family resilience framework
|
Aim: Explore and analyze the family resilience of patients with gynecological malignancies after radical hysterectomy, providing a theoretical basis for the formulation of future intervention measures.Methods: Using a phenomenological descriptive qualitative research method, 17 patients who underwent radical surgery for gynecological malignancies were selected for semi-structured interviews. Data analysis and theme extraction were conducted using Colaizzi data analysis method and NVivo V.12.Results: Three themes and eight sub-themes were extracted: family belief system (confront surgical challenges head-on, attribute positive significance to adversity, stay positive), family organization model (timely adjustment of family roles, family cohesion, get support and help from others), and family communication and problem solving skills (communicate to eliminate negative emotions, collaborative problem solving).Conclusion: This study indicates that the family belief system is the solid foundation of family resilience, the family organizational pattern serves as a buffer when the family faces adversity, and positive communication and collaborative problem solving create a positive feedback loop that enhances family resilience. Future interventions could enhance patients’ family resilience from the perspective of family strengths.
|
2234943X
|
ONCOLOGY
|
10.3390/ejihpe15050065
|
Predicting Workplace Hazard, Stress and Burnout Among Public Health Inspectors: An AI-Driven Analysis in the Context of Climate Change
|
The increasing severity of climate-related workplace hazards challenges occupational health and safety, particularly for Public Health and Safety Inspectors. Exposure to extreme temperatures, air pollution, and high-risk environments heightens immediate physical threats and long-term burnout. This study employs Artificial Intelligence (AI)-driven predictive analytics and secondary data analysis to assess hazards and forecast burnout risks. Machine learning models, including eXtreme Gradient Boosting (XGBoost 3.0), Random Forest, Autoencoders, and Long Short-Term Memory (LSTMs), achieved 85–90% accuracy in hazard prediction, reducing workplace incidents by 35% over six months. Burnout risk analysis identified key predictors: physical hazard exposure (β = 0.76, p < 0.01), extended work hours (>10 h/day, +40% risk), and inadequate training (β = 0.68, p < 0.05). Adaptive workload scheduling and fatigue monitoring reduced burnout prevalence by 28%. Real-time environmental data improved hazard detection, while Natural Language Processing (NLP)-based text mining identified stress-related indicators in worker reports. The results demonstrate AI’s effectiveness in workplace safety, predicting, classifying, and mitigating risks. Reinforcement learning-based adaptive monitoring optimizes workforce well-being. Expanding predictive-driven occupational health frameworks to broader industries could enhance safety protocols, ensuring proactive risk mitigation. Future applications include integrating biometric wearables and real-time physiological monitoring to improve predictive accuracy and strengthen occupational resilience.
|
22549625
|
PSYCHOLOGY
|
10.1186/s40359-025-02743-8
|
The relationship of fear of pain, pain anxiety, and fear-avoidance beliefs with perceived stress in surgical patients with postoperative kinesiophobia
|
Kinesiophobia is one of the most prevalent postoperative problems with negative effects on patient mobility. Fear of pain (FOP), pain anxiety (PA), and fear of avoidance beliefs (FABs) are influential factors on postoperative mobility and may be affected by perceived stress (PS). The present study examined whether perceived stress serves to mediate the relationship between fear of pain, pain anxiety, and fear of avoidance beliefs with kinesiophobia (fear of movement) in postoperative patients. The study was conducted in the neurosurgery, general surgery, and orthopedic wards of a hospital in Amol, Iran. A total of 330 patients (178 men and 152 women), aged 18 to 74 years, who had undergone various surgical procedures, were included. Participants were recruited using a consecutive sampling technique over a defined period to account for the staggered timing of surgeries and ensure broader representation. All patients were assessed six hours post-surgery using validated instruments, including the Tampa Scale for Kinesiophobia, Pain Anxiety Symptoms Scale, Fear of Pain Questionnaire, Fear-Avoidance Beliefs Questionnaire, and Perceived Stress Scale. The majority of the sample were men (53.9%), married (80%), with a mean age of 44.38 (SD = 13.49) years. Of the participants, 119 (36.1%) underwent orthopedic surgery, 139 (42.1%) underwent abdominal surgery, and 72 (21.8%) underwent surgery for discopathy. The path analysis revealed that kinesiophobia exhibited a significant relationship with FABs (β = 0.206; p < 0.001; 95% CI: 0.009 to 0.017) and PA (β = 0.474; p < 0.001; 95% CI: 0.021 to 0.031), while no significant relationship was found with FOP (β = 0.072; p = 0.408; 95% CI: -0.011 to 0.011). Also, the findings indicated that PS as mediator had a significant relationship with FABs (ß = 0.191; P < 0.001; 95% CI: 0.009 to 0.017), PA (ß = 0.393; P < 0.001;95% CI: 0.021 to 0.031), and kinesiophobia (ß = 0.812; P < 0.001; 95% CI: 0.021 to 0.031. The study found that pain anxiety and fear-avoidance beliefs are key factors contributing to kinesiophobia after surgery. Addressing these fears is important for improving postoperative mobility. Perceived stress mediated the relationship between these factors and kinesiophobia. Managing stress may be a helpful intervention to improve outcomes for postoperative patients. Healthcare providers should assess and address psychological factors like pain anxiety and fear-avoidance beliefs to promote better recovery and mobility in patients after surgery.
|
20507283
|
PSYCHOLOGY
|
10.1186/s40594-025-00545-3
|
Processes, challenges, and teacher roles in developing and implementing collaborative STEM curricula: case studies of two Taiwanese schools
|
A key research gap in current STEM education lies in the need for a more in-depth exploration of STEM teachers as curriculum designers, particularly in how they collaborate in designing STEM curricula and their roles within that process. This study selected two high-performing STEM teaching teams for investigation and employed a naturalistic approach along with a case study methodology to examine how STEM teachers collaborate to develop and implement STEM curricula in real teaching contexts. After 8 months of data collection and analysis, the main findings are as follows: (1) there are discrepancies between the tasks emphasized at each stage of the collaborative STEM curriculum model by high-performing STEM teaching teams and those outlined in theoretical models. In addition, the resources and drivers valued by these teams are not accounted for in the theoretical models. (2) Both high-performing STEM teaching teams faced several challenges during collaborative curriculum design and implementation, including difficulties with scheduling, limited time for lesson preparation, challenges in assessing higher-order thinking, and integrating team members. The main challenge faced by both schools was the absence of common meeting times for interdisciplinary collaboration. This highlights the need for strategic scheduling and institutional support to enable teacher collaboration in STEM education. (3) The three main roles within STEM teaching teams are leaders, core teachers, and participating teachers. However, in practice, core teachers and participating teachers often do not fulfill the responsibilities they are expected to undertake. This study also discusses potential research limitations and offers relevant suggestions for future research. The study also identified a significant discrepancy between theory and practice. While the PADPIE model outlines a structured six-stage design stages, schools frequently skip or merge stages due to time and resource limitations. An inconsistency was noted in the enactment of teacher roles. While formal assignments such as leaders, core teachers, and participating teachers were established, many core and participating teachers often lacked clarity and initiative in their responsibilities. These findings highlight the need to bridge the gap between theoretical models and real-world implementation.
|
21967822
|
EDUCATION
|
10.3390/ai6050088
|
Evaluating the Efficacy of Deep Learning Models for Identifying Manipulated Medical Fundus Images
|
(1) Background: The misuse of transformation technology using medical images is a critical problem that can endanger patients’ lives, and detecting manipulation via a deep learning model is essential to address issues of manipulated medical images that may arise in the healthcare field. (2) Methods: The dataset was divided into a real fundus dataset and a manipulated dataset. The fundus image manipulation detection model uses a deep learning model based on a Convolution Neural Network (CNN) structure that applies a concatenate operation for fast computation speed and reduced loss of input image weights. (3) Results: For real data, the model achieved an average sensitivity of 0.98, precision of 1.00, F1-score of 0.99, and AUC of 0.988. For manipulated data, the model recorded sensitivity of 1.00, precision of 0.84, F1-score of 0.92, and AUC of 0.988. Comparatively, five ophthalmologists achieved lower average scores on manipulated data: sensitivity of 0.71, precision of 0.61, F1-score of 0.65, and AUC of 0.822. (4) Conclusions: This study presents the possibility of addressing and preventing problems caused by manipulated medical images in the healthcare field. The proposed approach for detecting manipulated fundus images through a deep learning model demonstrates higher performance than that of ophthalmologists, making it an effective method.
|
26732688
|
AI
|
10.3389/fpsyg.2025.1491265
|
Impact of microlearning on developing soft skills of university students across disciplines
|
Introduction: This study explores the effectiveness of microlearning in developing key soft skills among university students across four academic disciplines: humanities and arts (HA), business studies (BS), medical sciences (MS), and technical and engineering (TE). Addressing the disconnect between academic training and industry expectations, the research investigates how microlearning interventions influence the development of teamwork skills (TWS), leadership skills (LS), communication skills (CS), time management skills (TMS), and emotional intelligence (EI). The study also aims to identify which disciplines benefit most from microlearning for each specific skill.Methods: A total of 384 Chinese university students participated in this study, with a questionnaire recovery rate of 93.23% and near-equal representation from each discipline. Participants completed pre- and post-intervention surveys following tailored microlearning modules. Statistical analyses—including paired sample t-tests, independent sample t-tests, and effect size calculations—were employed to test five hypotheses related to soft skill development across disciplines.Results: Findings indicate that leadership-focused microlearning modules significantly benefited TE and MS students, while EI training was particularly effective for BS students. Notable improvements in CS and TMS were observed among BS and TE students, aligning with skills demanded in corporate project management. Overall, microlearning interventions produced measurable enhancements in specific soft skills, with variation across academic disciplines.Discussion: The results suggest that integrating structured, discipline-specific microlearning into university curricula can effectively bridge academic-industry skill gaps. Faculty are encouraged to adopt scenario-based microlearning strategies to enhance student engagement. Higher education institutions should prioritize microlearning experience in student development and recruitment. Additionally, EdTech providers are urged to develop AI-powered interactive platforms to personalize learning, while students should proactively engage in targeted microlearning to improve academic and career outcomes.
|
16641078
|
PSYCHOLOGY
|
10.3389/fpsyg.2025.1579787
|
If not police, then who? Building a new workforce for community behavioral health crisis response
|
Introduction: Communities across the United States and elsewhere are working to implement alternatives to law enforcement as primary responders to behavioral health crises. These efforts can only be successful if there is a skilled workforce prepared to take on this role. We argue that this workforce must be developed, and propose a new, credentialed Community Behavioral Health Crisis Responder (CBHCR) role.Methods: Guided by a 13-member advisory board with expertise across behavioral health, crisis services, and law enforcement, we conducted a literature review, key informant interviews, and focus groups to identify the foundational values, competencies, and skills for this proposed role.Results: Interview and focus group participants discussed desired characteristics of CBHCRs and emphasized values such as cultural humility, a nonjudgmental approach, and the importance of lived experience broadly defined. Competencies and skills included engagement and communication strategies that enhance safety and trust, suicide prevention, conflict resolution, and situational awareness. Participants highlighted the need to train CBHCRs to provide compassionate, trauma-informed crisis intervention, de-escalation, support, and connection to needed resources. In conjunction with our advisory board and external experts, we used the findings to iteratively refine the values, competencies, and skills of CBHCRs.Discussion: We discuss the next steps in creating this new, skilled and credentialed crisis response workforce.
|
16641078
|
PSYCHOLOGY
|
10.3390/ai6050091
|
A Hybrid and Modular Integration Concept for Anomaly Detection in Industrial Control Systems
|
Effective anomaly detection is essential for realizing modern and secure industrial control systems. However, the direct integration of anomaly detection within such a system is complex due to the wide variety of hardware used, different communication protocols, and given industrial requirements. Many components of an industrial control system allow direct integration, while others are designed as closed systems or do not have the required performance. At the same time, the effective usage of available resources and the sustainable use of energy are more important than ever for modern industry. Therefore, in this paper, we present a modular and hybrid concept that enables the integration of efficient and effective anomaly detection while optimising the use of available resources under consideration of industrial requirements. Because of the modular and hybrid properties, many functionalities can be outsourced to the respective devices, and at the same time, additional hardware can be integrated where required. The resulting flexibility allows the seamless integration of complete anomaly detection into existing and legacy systems without the need for expensive centralised or cloud-based solutions. Through a detailed evaluation within an industrial unit, we demonstrate the performance and versatility of our concept.
|
26732688
|
AI
|
10.1007/s00432-025-06210-0
|
Clinical pathological characteristics and prognostic analysis of renal primitive neuroectodermal tumours: a multicentre retrospective study of 16 cases in Northwest China
|
Objective Renal primitive neuroectodermal tumours (rPNETs) are extremely rare and highly aggressive malignancy, posing significant diagnostic and therapeutic challenges. This study aims to describe the clinicopathological characteristics, treatment strategies, and survival outcomes of 16 cases of rPNET from multiple centers in Northwest China, and to explore potential prognostic factors. Methods A multicenter retrospective study was conducted, including 16 patients diagnosed with rPNET across five hospitals in Northwest China. Immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) were employed to assess the expression of molecular markers, including P53, BCL-2, Ki-67, and EWSR1 gene rearrangements. Survival analysis was performed using the Kaplan-Meier method, and prognostic factors were evaluated using univariate and multivariate Cox regression models. Results The median age of the patients was 39 years, with a median Ki-67 proliferation index of 50%. P53 mutations were detected in 87.0% of cases, and BCL-2 positive expression was observed in 56.25% of cases. The median overall survival (OS) was 14 months. Univariate analysis revealed that age, tumor stage, BCL-2 expression, and Ki-67 index were significantly associated with OS. Multivariate analysis identified high Ki-67 expression (HR = 1.100, 95% CI: 1.030–1.174, p = 0.004) and negative BCL-2 expression (HR = 0.151, 95% CI: 0.026–0.888, p = 0.037) as independent risk factors for poor prognosis. Kaplan-Meier survival curves demonstrated that the median OS was significantly shorter in patients with high Ki-67 expression (12 months) compared to those with low Ki-67 expression (20 months) (Log-rank test, P < 0.01). Similarly, the median OS was significantly shorter in the BCL-2 negative group (10 months) compared to the BCL-2 positive group (24 months) (Log-rank test, P < 0.05). Conclusion The absence of rosette structures does not exclude the diagnosis of rPNET. BCL-2 and Ki-67 expression are significant prognostic factors, with high Ki-67 expression and negative BCL-2 expression associated with worse outcomes. These findings highlight the importance of molecular markers in risk stratification and treatment planning for rPNET.
|
14321335
|
ONCOLOGY
|
10.3389/fonc.2025.1507677
|
Copy number gain of MET gene with low level in a metastatic lung adenocarcinoma patient represents response to salvage treatment with savolitinib and osimertinib: a case report
|
Background: Mesenchymal–epithelial transition (MET) amplification is one of the molecular mechanisms of abnormal MET oncogenic signaling in non-small cell lung cancer (NSCLC), significantly contributing to tumor cell survival, proliferation, metastasis, and drug resistance. The results of the TATTON trial showed that the combination of savolitinib and osimertinib can prolong the survival of patients with advanced EGFR-TKI-resistant NSCLC and high-level acquired MET amplification.Case presentation: We present a case of an NSCLC patient who exhibited acquired MET amplification with a gene copy number (GCN) of 3 following resistance to EGFR-TKI. The patient achieved a substantial response to salvage therapy with savolitinib and osimertinib, resulting in a 7-month progression-free survival (PFS).Conclusions: We considered that a regimen of savolitinib + osimertinib combination sometimes may still be potentially beneficial for NSCLC patients with low-GCN-level MET amplification. However, it needs further confirmation in a larger cohort.
|
2234943X
|
ONCOLOGY
|
10.3389/fonc.2025.1502062
|
Radiomics model based on dual-energy CT venous phase parameters to predict Ki-67 levels in gastrointestinal stromal tumors
|
Objective: To develop and validate a radiomics model based on the features of the Dual-Energy CT (DECT) venous phase iodine density maps and effective atomic number maps to predict Ki-67 expression levels in gastrointestinal stromal tumors (GISTs).Methods: A total of 91 patients with GIST were retrospectively analyzed, including 69 patients with low Ki-67 expression (≤5%) and 22 patients with high Ki-67 expression (>5%). Four clinical features (gender, age, maximum tumor diameter, and tumor location) were extracted to construct a clinical model. The venous phase enhanced CT iodine density maps and effective atomic number maps of DSCT were used to build radiomics models. Logistic regression was used to combine radiomics features with clinical features to build a combined model. Finally, the optimal model’s discrimination, calibration, and clinical decision curve were validated using the Bootstrap method.Results: The combined model was identified as the best model, with high predictive performance. The model’s discrimination had an AUC of 0.982 (95% CI, 0.9603-1). The calibration test showed a Hosmer-Lemeshow test P-value of 0.99. The clinical decision curve demonstrated a probability threshold range of 15% to 98%, with a high net benefit.Conclusion: The nomogram model combining clinical features and radiomics (iodine density map radscore + effective atomic number map radscore) has the highest accuracy for preoperative prediction of Ki-67 expression in GISTs.
|
2234943X
|
ONCOLOGY
|
10.3390/ai6050093
|
Personalized Non-Player Characters: A Framework for Character-Consistent Dialogue Generation
|
Generating character-consistent and personalized dialogue for Non-Player Characters (NPCs) in Role-Playing Games (RPGs) poses significant challenges, especially due to limited memory retention and inconsistent character representation. This paper proposes a framework for generating personalized dialogues based on character-specific knowledge. By combining static knowledge fine-tuning and dynamic knowledge graph technology, the framework generates dialogue content that is more aligned with character settings and is highly personalized. Specifically, the paper introduces a protective static knowledge fine-tuning approach to ensure that the language model does not generate content beyond the character’s cognitive scope during conversations. Additionally, dynamic knowledge graphs are employed to store and update the interaction history between NPCs and players, forming unique “experience-response” patterns. During dialogue generation, the paper first parses player input into an Abstract Meaning Representation (AMR) graph, retrieves relevant memory nodes from the knowledge graph, and constructs a fused graph structure. This integrated graph is encoded via a graph neural network to generate high-dimensional semantic vectors, which are then used to retrieve and supplement knowledge from the vector database. Ultimately, the model generates personalized responses consistent with the NPC’s identity. Experimental results demonstrate that the framework significantly enhances the authenticity of NPC dialogues and player immersion and performs well on multiple large-scale language models.
|
26732688
|
AI
|
10.3389/frai.2025.1529814
|
Precision enhancement in wireless capsule endoscopy: a novel transformer-based approach for real-time video object detection
|
Background: Wireless Capsule Endoscopy (WCE) enables non-invasive imaging of the gastrointestinal tract but generates vast video data, making real-time and accurate abnormality detection challenging. Traditional detection methods struggle with uncontrolled illumination, complex textures, and high-speed processing demands.Methods: This study presents a novel approach using Real-Time Detection Transformer (RT-DETR), a transformer-based object detection model, specifically optimized for WCE video analysis. The model captures contextual information between frames and handles variable image conditions. It was evaluated using the Kvasir-Capsule dataset, with performance assessed across three RT-DETR variants: Small (S), Medium (M), and X-Large (X).Results: RT-DETR-X achieved the highest detection precision. RT-DETR-M offered a practical trade-off between accuracy and speed, while RT-DETR-S processed frames at 270 FPS, enabling real-time performance. All three models demonstrated improved detection accuracy and computational efficiency compared to baseline methods.Discussion: The RT-DETR framework significantly enhances precision and real-time performance in gastrointestinal abnormality detection using WCE. Its clinical potential lies in supporting faster and more accurate diagnosis. Future work will focus on further optimization and deployment in endoscopic video analysis systems.
|
26248212
|
AI
|
10.3390/ai6050094
|
Robust Single-Cell RNA-Seq Analysis Using Hyperdimensional Computing: Enhanced Clustering and Classification Methods
|
Background. Single-cell RNA sequencing (scRNA-seq) has transformed genomics by enabling the study of cellular heterogeneity. However, its high dimensionality, noise, and sparsity pose significant challenges for data analysis. Methods. We investigate the use of Hyperdimensional Computing (HDC), a brain-inspired computational framework recognized for its noise robustness and hardware efficiency, to tackle the challenges in scRNA-seq data analysis. We apply HDC to both supervised classification and unsupervised clustering tasks. Results. Our experiments demonstrate that HDC consistently outperforms established methods such as XGBoost, Seurat reference mapping, and scANVI in terms of noise tolerance and scalability. HDC achieves superior accuracy in classification tasks and maintains robust clustering performance across varying noise levels. Conclusions. These results highlight HDC as a promising framework for accurate and efficient single-cell data analysis. Its potential extends to other high-dimensional biological datasets including proteomics, epigenomics, and transcriptomics, with implications for advancing bioinformatics and personalized medicine.
|
26732688
|
AI
|
10.3390/ejihpe15050069
|
Effect of an Educational Intervention on Pupil’s Knowledge, Attitudes, Perceptions, and Behavior on Air Pollution in Public Schools in Pristina
|
This interventional study aimed to assess the effectiveness of a school-based environmental education program on improving knowledge, attitudes, perceptions, and behavior related to air pollution among pupils in low-middle schools in Pristina, Kosovo. Air pollution is a pressing issue in Kosovo, particularly in urban areas, making it essential to raise awareness from an early age. As one of the first initiatives of its kind in the country, this study offers valuable insights into the impact of educational interventions on students’ understanding of environmental issues. The study involved an intervention group of fifth to ninth grade students who participated in a structured environmental education program, with data collected through pre-test, post-test, and follow-up assessment. We used a quantitative questionnaire with four sections—demographics, knowledge, perceptions, attitudes, and behavior. The findings revealed a significant improvement in knowledge and perceptions about air pollution among students in the intervention group, highlighting the crucial role of education in raising environmental awareness. However, the intervention had limited impact on changing attitudes and no significant effect on pro-environmental behavior, echoing challenges found in previous studies. Parental education, particularly maternal education, was found to play a substantial role in shaping attitudes, while gender and parental education positively influenced perceptions. The study also identified a negative association between higher grade levels and both knowledge and perception scores. Despite its success in enhancing knowledge, the short intervention period and challenges in participant engagement limited the program’s ability to drive long-term behavioral change. These findings emphasize the need for more sustained and comprehensive interventions to address the complex relationship between knowledge, attitudes, and environmental behaviors.
|
22549625
|
PSYCHOLOGY
|
10.3390/ai6050096
|
Automated Pruning Framework for Large Language Models Using Combinatorial Optimization
|
Currently, large language models (LLMs) have been utilized in many aspects of natural language processing. However, due to their significant size and high computational demands, large computational resources are required for deployment. In this research, we focus on the automated approach for size reduction of such a model. We propose the framework to perform the automated pruning based on combinatorial optimization. Two techniques were particularly studied, i.e., particle swarm optimization (PSO) and whale optimization algorithm (WOA). The model pruning problem was modeled as a combinatorial optimization task whose the goal is to minimize model size while maintaining model accuracy. The framework systematically explores the search space to identify the most optimal pruning configurations, removing redundant or non-contributory parameters. The two optimizations, PSO and WOA, were evaluated for their ability to efficiently navigate the search space. As a result, with PSO, the proposed framework can reduce the model size of Llama-3.1-70B by 13.44% while keeping the loss of model accuracy at 19.25%; with WOA, the model size reduction is 12.07% with 22.81% loss of model accuracy. Since accuracy degradation may occur during pruning process, the framework integrates the post-process to recover the model accuracy. After this process, the pruned model loss can reduce to 12.72% and 14.83% using PSO and WOA, respectively.
|
26732688
|
AI
|
10.3390/ejihpe15050070
|
The Influence of Loneliness, Social Support and Income on Mental Well-Being
|
Mental well-being is a multifaceted concept that reflects emotional stability, psychological resilience and social connectedness. This study examines how demographic factors, perceived loneliness, and social support influence mental well-being in Spain. Participants were surveyed online and provided personal information along with responses to the University of California, Los Angeles (UCLA) Loneliness Scale, the Medical Outcomes Study Social Support Survey (MOS-SSS), and the Warwick–Edinburgh Mental Well-Being Scale (WEMWBS). Our findings support previous research on mental well-being in Spain and again show significant associations between income, loneliness, social support and overall mental health. In particular, perceived loneliness was found to be a strong predictor of mental well-being. Furthermore, income and social support were found to partially mediate the relationship between loneliness and mental well-being. These findings highlight the critical role of social connections and financial stability in promoting mental health. Overall, this research contributes to the growing understanding of the factors influencing mental well-being and provides valuable insights for improving mental health outcomes.
|
22549625
|
PSYCHOLOGY
|
10.1007/s44196-025-00827-2
|
Min3GISG: A Synergistic Feature Selection Framework for Industrial Control System Security with the Integrating Genetic Algorithm and Filter Methods
|
Industrial control systems (ICS) are crucial for automating and optimizing industrial operations but are increasingly vulnerable to cyberattacks due to their interconnected nature. High-dimensional ICS datasets pose challenges for effective anomaly detection and classification. This study aims to enhance ICS security by improving attack detection through an optimized feature selection framework that balances dimensionality reduction and classification accuracy. The study utilizes the HAI dataset, comprising 54,000 time series records with 225 features representing normal and anomalous ICS behaviors. A hybrid feature selection approach integrating wrapper and filter methods was employed. Initially, a Genetic Algorithm (GA) identified 118 relevant features. Further refinement was conducted using filter-based methods—Symmetrical Uncertainty (SU), Information Gain (IG), and Gain Ratio (GR)—leading to a final subset of 104 optimal features. These features were used to train classification models (Naive Bayes (NB), Random Forest (RF), and Support Vector Machine (SVM)) with a 70:30 train-test split and tenfold cross-validation. The proposed feature selection method significantly improved classification accuracy, achieving 98.86% (NB), 99.91% (RF), and 97.97% (SVM). Compared to the full dataset (225 features), which yielded 97.51%, 99.93%, and 96.17%, respectively, our optimized feature subset maintained or enhanced classification performance while reducing computational complexity. This research demonstrates the effectiveness of a hybrid feature selection approach in improving ICS anomaly detection. By reducing feature dimensionality without compromising accuracy, the proposed method enhances ICS security, offering a scalable and efficient solution for real-time attack detection.
|
18756883
|
AI
|
10.3389/fpsyg.2025.1491759
|
The Illusory Health Beliefs Scale: validation using exploratory structural equation modeling and multidimensional Rasch analysis
|
The Illusory Health Beliefs Scale (IHBS) is a multidimensional instrument that evaluates endorsement of scientifically unsubstantiated, illusory health-oriented notions. These beliefs are important because they potentially influence attitudes/actions to the detriment of personal wellbeing/health. Preceding research examining IHBS item performance at the unidimensional subscale level identified five dimensions (Religious/Spiritual, Superstition, Precognitive, Health Myths, Skepticism), and an independent Health Pseudoscience subscale. The present paper extended latent structure analysis by employing exploratory structural equation modeling (ESEM) and multidimensional Rasch analysis. Concurrently, statistical appraisal tested convergent validity via relationships with related health-based constructs (i.e., health locus of control, HLC and beliefs about complementary and alternative medicine, CAM). A sample of 2,138 completed the IHBS (1,016 males, 1,113 females, seven non-binary, two preferred not to disclose). Following minor scale modification, ESEM reported good data-fit for a six-factor model. With the exception of Skepticism, which was negatively associated, IHBS subfactors correlated positively with HLC and CAM. These outcomes supported the supposition that the IHBS measures perceived and illusory health control. Rasch analysis designated sufficient multidimensionality and satisfactory subscale functioning. Strong associations indicated that IHBS dimensions assessed related but discrete aspects of illusory health beliefs. High associations among paranormal-based dimensions (Religious/Spiritual, Superstition, and Precognitive) suggested the need for greater content separation. Moreover, the poor reliability of Skepticism designated the need to develop a more efficacious assessment of this dimension.
|
16641078
|
PSYCHOLOGY
|
10.3390/cancers17101613
|
Molecular Mechanisms of Drug Resistance in Clear Cell Renal Cell Carcinoma
|
Renal cell carcinoma (RCC) accounts for about 3% of all human tumors. Alterations of oxygen, lipids, iron, and energy metabolism are involved in carcinogenesis, development, and expansion. Thirty percent of patients affected by clear cell renal cell carcinoma (ccRCC) will develop relapses or distance metastases (mRCC), dramatically reducing their life expectancy. Current first-line therapies for mRCC patients are based on treatment with immune checkpoint inhibitors (ICIs) alone and in combination with each other or with tyrosine kinase inhibitors (TKIs). However, only 20% of patients show a mild response because of innate or acquired drug resistance during long-term treatment; therefore, resistant patients need alternative first-line or second-line therapies. Pharmacological resistance represents a big problem that counteracts the efficacy of treatment by reducing overall survival (OS) in mRCC patients. Investigating the molecular mechanisms underlying drug resistance is crucial to overcoming drug insensitivity and enhancing therapeutic outcomes. In this review, we emphasize the latest and most significant studies on the molecular mechanisms that drive drug resistance in ccRCC carcinoma. Particular attention is given to the key signaling pathways involved in resistance, including those mediated by HIF, p53, Akt-mTOR, MEK–ERK cascades, Wnt signaling, autophagy, membrane transporters, ferroptosis, and non-coding RNAs. Understanding these resistance mechanisms is essential for developing new therapeutic strategies aimed to enhancing overall OS and improving the quality of life for mRCC patients. This review also discusses recent clinical trial findings on the use of specific inhibitors able to circumvent drug resistance. The data presented here could be valuable for clinicians in understanding the mechanisms of drug resistance, ultimately aiding in the management of ccRCC patients.
|
20726694
|
ONCOLOGY
|
10.1007/s00432-025-06218-6
|
Thoracoscopy-guided thoracic paravertebral block using dexmedetomidine in combination with ropivacaine for postoperative analgesia after thoracoscopic radical resection of lung cancer: a randomized controlled trial
|
Purpose The aim of this trial was to evaluate the analgesic effect of dexmedetomidine combined with ropivacaine for thoracoscopic-guided thoracic paravertebral block (TTPB) after thoracoscopic radical resection (TRR) of lung cancer. Methods A total of 60 patients were enrolled from our hospital who underwent elective TRR of lung cancer and randomized into either a control group (group C) or a dexmedetomidine group (group D). Prior to incisional suturing, group C received ropivacaine alone for TTPB, while group D received dexmedetomidine combined with ropivacaine for TTPB. The primary outcome was the time to the first analgesic request (TFAR). The secondary outcomes included heart rate (HR), mean arterial pressure (MAP), Ramsay sedation score, and Numerical Rating Scale (NRS) scores (both at rest and during coughing) at the following time points: before the TTPB operation (T0), 1 h postoperatively (T1), 2 h postoperatively (T2), 6 h postoperatively (T3), 12 h postoperatively (T4), 24 h postoperatively (T5), as well as 48 h postoperatively (T6). Additional secondary outcomes included the patient-controlled intravenous analgesia (PCIA) sufentanil dosage at 48 h postoperatively, the incidence of adverse reactions, and postoperative recovery. Results Compared to group C, group D showed a longer TFAR, lower total PCIA sufentanil dosage at 48 h postoperatively, and lower NRS scores at all time points; Group D also had lower MAP and HR, higher Ramsay sedation scores from T1 to T3 after surgery, a higher incidence of drowsiness, and better postoperative recovery. Conclusions As an adjuvant in combination with ropivacaine, dexmedetomidine enhanced the analgesic effect of TTPB, prolonged the duration of analgesia, and reduced the time to first ambulation and hospital stay. Clinical Trial Registration ChiCTR2400086347, Registered 28/06/2024.
|
14321335
|
ONCOLOGY
|
10.3389/feduc.2025.1596635
|
Financial literacy and educational level in Ecuadorian students: a structural analysis
|
Background: Financial literacy has been recognized as a key competency for making in-formed economic decisions, particularly in contexts where access to financial products exceeds the population’s literacy level. However, in Ecuador, persistent gaps remain be-tween formal educational attainment and applied financial knowledge. In this context, the objective of this study was to analyze the relationship between educational level and financial literacy among Ecuadorian students.Methods: A quantitative approach was adopted, with a descriptive-correlational level, non-experimental type, and cross-sectional design. The sample consisted of 2,021 participants, selected through non-probabilistic convenience sampling. A structured questionnaire of 33 items was administered, distributed across four analytical dimensions. Statistical analysis was performed using SPSS and AMOS, including reliability testing, factorial validity, and structural model fit.Results: The results revealed that educational level has a significant effect on financial literacy. Individuals with higher education exhibited the highest levels, while those who completed only primary education showed the lowest. Four latent factors were validated: technical knowledge, socioeconomic impact of financial education, practical application of knowledge, and financial self-management.Conclusion: The correlations between these factors were strong and statistically significant, highlighting the pivotal role of educational level in shaping financial literacy. The proposed model presents a valid and consistent structure, effectively reflecting the relationships between the key variables. These findings emphasize the necessity for tailored and context-specific educational interventions that address the diverse needs of different population segments, thereby enhancing financial literacy across varying educational levels.
|
2504284X
|
EDUCATION
|
10.3390/educsci15050593
|
Everyone Is Reading and Playing! A Participatory Theatre Project to Promote Reading Competence
|
This study explores the use of a theatre project to enhance reading competencies among students with special educational needs (SENs) in inclusive classrooms. The project, titled “Stop Bullying! A Theatre Project”, aimed to improve students’ reading skills through dramatised engagement with texts, with a particular focus on promoting literacy and social interaction. Employing a Design-Based Research (DBR) methodology, the study involved iterative cycles of implementation and data collection. Participants, including students with varying reading abilities, engaged in theatrical activities that incorporated reading strategies such as reading aloud, paired reading, and choral reading—each designed to support comprehension, fluency, and reading confidence. Findings from multiple cycles indicated improvements in students’ social dynamics, including stronger peer interactions and increased group cohesion. While quantitative reading assessment data showed only modest gains in reading performance, qualitative observations revealed significant improvements in reading skills and social interactions during collaborative performances. The study concludes that a theatre-based approach can effectively support reading development while fostering a more inclusive and supportive classroom environment.
|
22277102
|
EDUCATION
|
10.3390/educsci15050603
|
Developing Elite Strength and Conditioning Coaches’ Practice Through Facilitated Reflection
|
Recent research has suggested that strength and conditioning (S&C) coach development should consider constructivist learning theories to promote coach development and learning of psychosocial coaching competencies. Reflective practice can encourage holistic learning through promoting an internal dialogue of the meaningfulness of an individual’s experiences. Our study aimed to examine the efficacy of a facilitated, guided, and longitudinal reflective process to promote coach learning of psychosocial coaching practice using Moon’s reflective framework. Over a four-week period, six elite S&C coaches engaged in a guided process reflection process with a facilitator. This included daily journaling in an e-diary with the facilitator providing feedback at the end of each week. At the end, each S&C coach participated in an exit interview. Data were analysed using interpretative phenomenological analysis. Findings revealed that there were potential benefits for the S&C coach’s process of reflection such as providing accountability through developing a close relationship with the facilitator, which enabled the S&C coaches to more critically link learning to behaviour change. Furthermore, S&C coaches’ learning resulted in developing awareness of self/athlete’s needs, increased flexibility, and enhanced confidence. This resulted in S&C coaches developing psychosocial coaching competencies that enabled them to change their practice to become more athlete centred.
|
22277102
|
EDUCATION
|
10.3389/feduc.2025.1574962
|
Using website creation as a hub, promoted collaborative learning in teacher education
|
Introduction: The article presents a teaching design used in the first year of a teacher education in Norway. The teacher educators designed, taught and researched the project, evaluating it in collaboration with students. This is in line with a practitioner/action research approach and formative dialogue research. The teaching design was centered around the making of websites as a form of wiki learning. 143 students participated in the project. This article focuses on the student perspective and the data material was gathered through a survey and reflection papers.Results: The students pointed out the collaboration as the most important learning and how that prepared them for the future teacher profession. The student statements were classified into three categories of professional growth and discussed using the theory of professional capital. Using the students’ comments and the teacher educators’ experience, the teaching design is discussed up against earlier designs of wiki learning. The discussion elaborates on 6 possible success factors.
|
2504284X
|
EDUCATION
|
10.3389/feduc.2025.1600497
|
Academic writing strategies in university students from three disciplinary areas: design and validation of an instrument
|
It is acknowledged that writing strategies are highly important not only for their usefulness in the process of text production but also for their value as a learning technique, as they involve the use of cognitive and metacognitive operations. Various international studies have validated scales to measure their use and have helped characterize expert and non-expert university writers. However, regarding specific strategies in Spanish as a mother tongue, further research is still needed in the context of higher education students. This study aimed to develop a standardized measurement instrument to identify university students' use of academic writing strategies according to disciplinary area. The sample consisted of 290 students from the Humanities and Social Sciences, Health Sciences, and Engineering from a regional university in Chile. A Likert-type scale was applied whose design was based on a thorough review of successful instruments from different parts of the world. The results indicate that the developed instrument is appropriate for higher education contexts and across different disciplinary areas.
|
2504284X
|
EDUCATION
|
10.3390/ejihpe15050086
|
Neuroscience Exposure as a Predictor of Teaching Self-Efficacy
|
Teaching self-efficacy refers to a teacher’s confidence in their ability to engage students and foster learning, directly influencing their instructional planning, strategies, and student assessment practices. Neuroscience education for teachers has been shown to increase enthusiasm and support professional growth by introducing essential brain-related principles. This study investigated whether prior exposure to neuroscience predicts teaching self-efficacy among Brazilian basic education teachers. A total of 1120 teachers completed online surveys, providing sociodemographic information, educational background, teaching experience, and data regarding their previous neuroscience exposure. Participants’ neuroscience knowledge was assessed through a questionnaire designed to measure familiarity with fundamental neuroscience concepts, and teaching self-efficacy was evaluated using the Teacher Sense of Efficacy Scale (TSES). The results indicated that teachers with prior exposure to extracurricular neuroscience courses demonstrated significantly higher neuroscience knowledge. Additionally, those with previous neuroscience exposure exhibited a marginally significant increase in self-efficacy for instructional strategies and a significant increase in classroom management, while no significant differences were observed in student engagement. Regression analyses confirmed that neuroscience exposure significantly predicted self-efficacy in instructional strategies and classroom management. These findings reinforce the connection between neuroscience education and enhanced teaching self-efficacy, underscoring the importance of neuroeducation programs as valuable tools for supporting teachers’ professional development and well-being.
|
22549625
|
PSYCHOLOGY
|
10.3390/ai6050102
|
Classification of Exoplanetary Light Curves Using Artificial Intelligence
|
In this article, we propose a robust star classification methodology leveraging light curves collected from 15 datasets within the Kepler field in the visible optical spectrum. By employing a Bagging neural network ensemble approach, specifically an Bagging-Performance Approach Neural Network (BAPANN), which integrates three supervised neural network architectures, we successfully classified 760 samples of curves which represent 9 type of stars. Our method demonstrated a high classification accuracy of up to 97% using light curve datasets containing 13, 20, 50, 150, and 450 points per star. The BAPANN achieved a minimum error rate of 0.1559 and exhibited efficient learning, requiring an average of 29 epochs. Additionally, nine types of stellar variability were classified through 45 conducted tests, taking into account error margins of 0, 5, and 10 for the light curve samples. These results highlight the BAPANN model’s robustness against uncertainty and ability to converge quickly in terms of iterations needed for learning, training, and validation.
|
26732688
|
AI
|
10.3389/frai.2025.1545607
|
Can chatbots teach us how to behave? Examining assumptions about user interactions with AI assistants and their social implications
|
In this article we examine the issue of AI assistants, and the way they respond to insults and sexually explicit requests. Public concern over these responses, particularly because AI assistants are usually female-voiced, prompted tech companies to make them more assertive. Researchers have explored whether these female-voiced AI assistants could encourage abusive behavior and reinforce societal sexism. However, the extent and nature of the problem are unclear due to a lack of data on user interactions. By combining psychological and socio-cultural perspectives, we problematize these assumptions and outline a number of research questions for leveraging AI assistants to promote gender inclusivity more effectively.
|
26248212
|
AI
|
10.1007/s44196-025-00848-x
|
A Novel Human Action Recognition Model by Grad-CAM Visualization with Multi-level Feature Extraction Using Global Average Pooling with Sequence Modeling by Bidirectional Gated Recurrent Units
|
Human action recognition is essential in many real-world scenarios, such as video surveillance, human–computer interaction, and behavior analysis. Despite the progress in deep learning, issues such as occlusion, distraction from the background, and motion pattern variability still exist, thus restricting the generalization ability of current models. Most methods are based only on spatial or temporal features and cannot efficiently capture both in one framework, causing lower accuracy in realistic situations. In response to these shortcomings, a multilevel feature extraction approach was proposed by integrating spatial and temporal features to improve the action recognition precision. The method captures RGB frames, optical flow, spatial saliency maps, and temporal saliency maps to enable an overall inspection of video streams. Efficient feature extraction was achieved by applying a pre-trained Inception V3 model and then bidirectional gated recurrent units (Bi-GRUs) to include sequential modeling. An attention mechanism was also included to boost the classification process by focusing on key temporal segments. UCF101 and HMDB51 benchmark datasets evaluated the efficiency of the strategy. The model’s accuracy was 98.13% on UCF101 and 81.45% on HMDB51, which validated the superior discrimination ability of the model in processing heterogeneous human actions. These results confirm that the provided framework is an efficient and discriminative action recognition approach, thus suitable for applications requiring extensive motion analysis and real-time deployment.
|
18756883
|
AI
|
10.3389/feduc.2025.1571711
|
Persistence patterns among secondary STEM teachers: a comparative study of Noyce scholar cohorts in face-to-face and blended learning environments amid the pandemic
|
Noyce scholars were provided funding to compete teaching certification in STEM and earn a master’s degree. Then, they were required to teach for 2 years in a Title I school setting. All cohorts were impacted by the pandemic (e.g., university coursework, student teaching and/or teaching was converted to blended learning). This study highlights the differences in teaching persistence across the three cohorts of scholars (n = 24) regarding continuance in earning their degrees and completing their two-year teaching obligation. Descriptive case study methodology was used in this comparative study across three cohorts. The primary research question explored how different modalities of initial teaching experiences impact early persistence among secondary STEM teachers. The supplemental research question explored scholars’ intention to remain in the teaching profession. Results indicated that the cohort with blended first year teaching experiences had the lowest persistence rate. Generally, scholars intend on persisting in the profession for 6 years or more. Recommendations for practice include the need for more traditional, face-to-face initial teaching experiences and a cohort model for new teachers. Recommendations for research include continued evaluation of Noyce projects, longitudinal studies to track STEM teachers’ persistence, and a comprehensive analysis of teacher preparation programs’ effectiveness in promoting teacher retention.
|
2504284X
|
EDUCATION
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.