12703036 research outputs found
Sort by
Fast-powerformer achieves accurate and memory-efficient mid-term wind power forecasting
No full textAbstract Wind power forecasting (WPF), as a significant research topic within renewable energy, plays a crucial role in enhancing the security, stability, and economic operation of power grids. However, mid-term forecasting faces a persistent dilemma: achieving high predictive accuracy often comes at the cost of computational efficiency. Existing Transformer-based architectures struggle with this trade-off: traditional temporal attention mechanisms suffer from computational redundancy and weak inter-variable coupling, while recent transposed architectures, despite improving speed, inherently compromise the capture of local temporal dynamics and domain-specific periodic characteristics. To overcome these limitations, this paper proposes Fast-Powerformer. Built upon the Reformer backbone, the model reconstructs the feature extraction paradigm through three complementary strategies: (1) an Input Transposition Mechanism that optimizes multivariate coupling modeling while reducing sequence complexity; (2) a lightweight temporal embedding module that compensates for the intrinsic deficiency of transposed architectures in capturing local sequential features; and (3) a Frequency Enhanced Channel Attention Mechanism (FECAM) that exploits spectral information to characterize the physical periodic patterns of wind power. Experimental results on multiple real-world wind farm datasets demonstrate that Fast-Powerformer achieves the best overall performance among compared methods. The model successfully balances superior accuracy with reduced resource consumption, highlighting its significant practical potential for resource-constrained scenarios
Global temperature anomaly prediction by using additive twin LSTM networks
No full textAbstract Due to the complexity of climate systems, data-driven modeling based on observed time series data is essential for predicting future climatic trends. This study aims to improve the long-term global temperature anomaly forecast performance of Long Short-Term Memory (LSTM) based neural network models. Although several LSTM variants and hybrid architectures have been suggested for time series data prediction problems, the long-term forecast performance of these models may not be satisfactory in practice. To address solution of these problems, firstly, authors focused on evaluating the forecast performance of models and suggested performance and test assessment procedures. Secondly, authors suggest an Additive Twin LSTM (AT-LSTM) model that can improve the forecast performance for the global temperature anomaly. Our test on the Berkeley Global Temperature Anomaly dataset demonstrates that the proposed AT-LSTM can improve performance relative to conventional LSTM variants in long-term forecasting. Authors observed that global temperature trend projections of the AT-LSTM models for 20 years in future are consistent with expectations of climate organizations and projections in other works. The AT-LSTM models forecasted an average of 1.415 °C with ± 0.073 °C error in the year 2042 and this indicates the strong potential of major climate changes in the near future of Earth
Characterisation of pelagic seascapes through micronektonic and zooplanktonic scattering layers
No full textAbstract Landscape ecology is a key discipline for studying the relationships between spatial patterns and ecological processes, as well as for monitoring macro-scale changes in ecosystems. Unlike terrestrial landscapes, which have been extensively studied, the open-ocean pelagic environment presents unique complexities that require innovative approaches for its understanding. Aggregations of micronektonic and macrozooplanktonic organisms in layers are prominent features of the open-ocean pelagic zone. Such visually cryptic features can be revealed by echosounders as pelagic Sound Scattering Layers (SSLs). We characterise various pelagic seascapes and their relationships with environmental parameters across three oceanographically contrasting tropical regions, characterised by diverse ecological patterns, using an integrated methodological framework that combines dual-frequency acoustic analysis (18 and 38 kHz). Diel vertical migration is a common feature that involves epipelagic and mesopelagic SSLs. Nevertheless, there are significant regional contrasts in SSL spatial distribution as oceanographic features influence SSL patterns and micronektonic acoustic backscatter. Acoustically defined pelagic seascapes reveal biological-physical coupling and SSL responses to oceanographic variability at meso- and macro-scales. SSL distribution was significantly driven by oceanographic variables such as temperature, chlorophyll a, salinity, oxygen, and PAR, as well as by mesoscale eddies that structured their spatial patterns, with anticyclonic eddies concentrating SSLs’ acoustic backscatter at their centres and cyclonic eddies exhibiting scattered acoustic backscatter at their peripheries. This framework enhances our ability to assess how climate variability and changing ocean conditions influence open-ocean pelagic ecosystems. Developed and demonstrated at a broad regional scale, this validated approach establishes a transferable framework for characterising pelagic habitats through integrated SSL structures, enabling its application across wider spatial and temporal domains to advance understanding of global biophysical and ecosystem dynamics
Comparison of artificial intelligence and multidisciplinary team recommendations in the management of colorectal cancer liver metastases
No full textAbstract Multidisciplinary teams (MDTs) are central to treatment planning for colorectal cancer liver metastases (CRCLM) but require time and consistent access to expertise. Chat-based large language models (LLMs) such as ChatGPT can generate recommendations from written clinical summaries; however, their concordance with MDT decisions in CRCLM is not well characterized. We conducted a single-center retrospective concordance study of 30 consecutive CRCLM cases discussed at an MDT. ChatGPT was provided a standardized anonymized text synopsis (without direct imaging access) and asked for management recommendations under two a priori conditions: (1) baseline synopsis only, and (2) a conditional query in which resectability status was explicitly specified. Each case and condition was queried three independent times in separate sessions using identical prompts; outputs were mapped to predefined management categories. Agreement between the final LLM recommendation and MDT decisions was assessed using percent agreement and Cohen’s kappa. Across repeated runs, the LLM assigned the same management category in all cases (within-model consistency 100%, 3/3) for both querying conditions. In the baseline condition, agreement with MDT decisions was 66.7% (20/30; Cohen’s kappa = 0.606, moderate agreement). In the conditional resectability-specified condition, agreement was 93.3% (28/30; Cohen’s kappa = 0.924, very good agreement). Baseline discordant cases were characterized by conservative model outputs, including recommendations for systemic therapy and/or additional diagnostic work-up; only two cases remained discordant after resectability was specified. A chat-based LLM showed moderate concordance with unanimous MDT recommendations from minimal case summaries and very good concordance when resectability status was explicitly specified. These findings support feasibility as a supervised decision-support adjunct, but do not establish clinical benefit; prospective outcome-based validation is required
An explainable hybrid CNN–transformer model for sign language recognition on edge devices using adaptive fusion and knowledge distillation
No full textAbstract Despite recent advances in deep learning (DL) for sign language recognition (SLR), most existing systems remain limited to monolingual datasets, lack interpretability, and are too computationally intensive for real-time edge deployment. With the growing need for inclusive and real-time communication technologies, efficient and deployable SLR systems are of critical importance. This paper presents TinyMSLR, an explainable, lightweight framework designed for isolated-sign (gloss) classification on resource-constrained devices. TinyMSLR combines a ConvNeXt-Tiny encoder for fine-grained local visual cues with a Swin Transformer encoder for long-range spatio-temporal context, and integrates an adaptive fusion gate to balance both streams. To further improve accuracy under strict compute and memory budgets, we introduce a dual-teacher knowledge distillation (KD) scheme that transfers complementary spatial and contextual knowledge from high-capacity CNN and Transformer teachers to the compact student model. We evaluate TinyMSLR in a controlled multilingual setting using two public datasets (DGS RWTH-PHOENIX-Weather 2014T and Mandarin CSL) by constructing a shared subset of 20 semantically aligned sign classes and segmenting RWTH continuous sequences into single-gloss clips. Therefore, all reported results correspond to isolated-sign recognition rather than continuous sentence-level multilingual CSLR. On this benchmark, TinyMSLR achieves 99.28% training accuracy and 99.01% validation accuracy, with an F1-score of 98.96%, while keeping the parameter count under 2.7M. Inference latency is 24 ms on standard CPUs and under 13.5 ms on edge GPUs. Overall, TinyMSLR demonstrates a practical accuracy–efficiency–explainability trade-off that is well aligned with deployment-ready multilingual isolated-sign systems on the edge
Design and validation of the perceptions of teacher authoritarianism in research scale (PTARS)
No full textAbstract Introduction Although certain factors limit students’ scientific production, one such factor, i.e., teacher authoritarianism during the preparation of research papers, has received little attention to date. Objective: To design and validate the Perceptions of Teacher Authoritarianism in Research Scale (PTARS). Method This research took the form of an instrumental study, in which 816 students and graduates from state and private universities located in the three regions of Peru participated voluntarily. The process by which the scale was constructed involved 10 stages, and the initial version of the scale consisted of 20 items. The data were subjected to descriptive analysis, exploratory factorial analysis (EFA), and confirmatory factor analysis (CFA), alongside tests of measurement invariance and reliability. Results The re-specification of the exploratory three-factor model, excluding items 7, 8, and 11 due to factorial complexity, significantly improved the fit indices. This was reflected in a reduction of the RMSEA from 0.274 to 0.067 and an increase of the CFI to 0.997. Subsequently, in the confirmatory factor analysis, the final 10-item model (Model 5) demonstrated excellent fit (χ² = 47.81, RMSEA = 0.02) and high internal consistency (ω ≥ 0.90 for all factors). In the invariance analysis, the configural model showed no differences exceeding the established cut-off points (ΔRMSEA < .015; ΔCFI ≤ .010), supporting the stability of the model across different gender groups. Conclusion This study represents an important step in the process of evaluating teacher authoritarianism during the preparation of research papers; therefore, the PTARS serves as a tool that, according to the results of this research, is both valid and reliable
Factors influencing suicidal ideation in Chinese adolescents with first-episode depressive disorder: a cross-sectional study
No full textAbstract Background Given the high prevalence and severe consequences of suicidal ideation (SI) in adolescents, it is critical to identify multidimensional predictors of the illness. This study aimed to investigate the physiological, psychological, and sociological factors that influence SI in adolescents with first-episode depressive disorder. Methods The study was recruited through convenience sampling. Data were collected using self-designed questionnaires, the Hamilton Depression Scale 24-item, the Hamilton Anxiety Scale, and the Simplified Coping Style Questionnaire. Thyroid function, cortisol, lipids, and event-related potential were measured in the participants. All independent variables were included in the model for logistic regression. Statistical analysis of data was done using SPSS version 25.0. Results The present study was for the inclusion of 150 adolescents who presented with first-episode depressive disorder, of whom 96 (64.00%) had SI. Females and older adolescents were more prone to SI. Depressive symptoms, anxiety symptoms, negative coping styles, total cholesterol levels, and the latency of P3a and P3b were positively correlated with SI. Positive coping style, N2 amplitude, and SI were negatively correlated. Conclusions SI in Chinese adolescents with first-episode depressive disorder demonstrates a multifactorial relationship, involving physiological, psychological, and sociological factors. It is necessary to conduct multidomain, comprehensive assessments and develop intervention strategies
The effect of mindfulness-based preoperative education on postoperative pain: a Solomon four-group randomized controlled trial
No full textAbstract Aim This study aimed to evaluate the effectiveness of mindfulness-based preoperative education on postoperative pain-related cognitive intrusion and fear of pain among patients undergoing elective inguinal hernia surgery. Methods A Solomon four-group randomized controlled trial design was employed to control for potential pretest effects. A total of 68 patients scheduled for mesh-based inguinal hernia repair were randomly assigned to two experimental groups (EG1, EG2) and two control groups (CG1, CG2). Pretests were administered only to EG1 and CG1. The intervention consisted of awareness-based mindfulness education delivered preoperatively. Data were collected using the Sociodemographic Information Form, the Experience of Cognitive Intrusion of Pain Scale (ECIPS), and the Fear of Pain Questionnaire-III (FPQ-III). Statistical analyses were conducted with nonparametric tests, and a p-value .05). Postoperative comparisons among the four groups revealed no statistically significant differences in cognitive intrusion of pain or fear-of-pain subdimensions. However, when pre–post changes were examined between EG1 and CG1, the mean ECIPS score decreased by 5.9 points in EG1 but increased by 10.2 points in CG1, a significant difference favoring the intervention group (p = 0.005, r = 0.49). No significant between-group differences were found for the FPQ-III subscales of severe, mild, or medical pain fear (p = 0.499, p = 0.690, p = 0.112). Conclusion The findings indicate that mindfulness-based education may be associated with reductions in postoperative cognitive intrusion of pain in pretest–posttest comparisons, although most outcomes did not reach statistical significance. Although changes in pain-related fear were not significant, the findings indicate that mindfulness-informed education may enhance patients’ psychological readiness before surgery. Future studies with larger and more diverse samples are recommended to confirm these findings and explore long-term effects. Trial Registration NCT06449144 (Registry date: 07 July 2024)
Psychological and technological predictors of AI literacy profiles: a latent profile analysis among Chinese college students
No full textAbstract Background AI literacy is increasingly important in college students’ academic achievement, daily life, and future employability. However, current research predominantly overlooks the heterogeneity in students’ AI literacy, especially how individual psychological characteristics and features of AI technology contribute to this variation. This oversight limits the formulation of tailored strategies to meet the students’ various demands in an era shaped by rapid AI advancement. Objectives This study aims to adopt an individual-centered approach to identify distinct AI literacy profiles among college students. In addition, it investigates, based on affordance theory, how positive emotions, instrumental motivation, perceived ease of use, and psychological anthropomorphism predict assignment to different profiles. Methods A total of 808 Chinese college students participated in this survey. Latent profile analysis (LPA) was employed to classify students into distinct AI literacy profiles. Multinomial logistic regression was conducted to examine how psychological and technological factors predict profile classification. Findings This study identified four distinct AI literacy profiles among college students: preliminary contact type, ethical orientation type, balanced development type, and behavioral conservatism type. These profiles showed significant differences in positive emotions, instrumental motivation, perceived ease of use, and psychological anthropomorphism, highlighting diverse psychological and technological characteristics inherent to each group. Conclusions This study underscores the heterogeneity of AI literacy within the college student population and detects four distinct AI literacy profiles with unique psychological and technological traits. The findings indicate that students’ AI literacy is profoundly affected by emotional tendencies, motivational drives, and technological variables, highlighting the need for tailored educational strategies that address the distinct psychological and technological drivers of each literacy profile
Examining the relationship between quest for significance and smartphone addiction: does the dual passion model mediate this relationship?
No full textAbstract Background In Turkey, smartphone use exceeds both European and global averages, posing potential risks to individuals’ physical, social, and psychological well-being. The present research aimed to examine the relationships among smartphone addiction, significance quest, and passion, as well as to identify the mediating role of passion in the relationship between significance quest and smartphone addiction. In this regard, since there was no instrument available to assess passion within the Turkish cultural context, Study 1 adapted the Passion Scale for use in Turkish culture. Study 2 then investigated the associations among smartphone addiction, significance quest, and passion. Methods During the adaptation process of the Passion Scale, SPSS 25.0 was used for the exploratory factor analysis (EFA), and Mplus 7 was employed for the confirmatory factor analysis (CFA). In Study 1, the measurement invariance of the Turkish version of the scale was also examined across groups formed according to gender, age and types of activities performed with passion. The sample sizes for EFA and CFA are 270 and 289, respectively. Majority of participants in both groups are aged between 19 and 25 (EFA: 71.2%; CFA: 66.1%). In Study 2, after testing multivariate statistical assumptions, the SEM-based mediation model was tested using Mplus 7. The study group of Study 2 consisted of 674 individuals from different ages (22.6% of participants were adolescents aged 15–18; 57.1% of participanst aged 19–25; 20.3% of participants aged 26–35). Results The findings of Study 1 demonstrated that the two-factor structure of the Passion Scale was confirmed within the Turkish cultural context and that the scale exhibited high reliability. In Study 2, positive and significant relationships were identified between significance quest and smartphone addiction, between significance quest and obsessive passion, and between obsessive passion and smartphone addiction. Moreover, obsessive passion was found to mediate the relationship between significance quest and smartphone addiction. In contrast, no significant associations were observed between significance quest and harmonic passion, nor between harmonic passion and smartphone addiction. Conclusions The results of the study suggest that obsessive passion is one of the factors that strengthens the relationship between the significance quest and smartphone addiction. When an individual with obsessive passion experiences a loss of significance, their usage process may progress toward smartphone addiction, even if they are not yet addicted
Data sources:
Access Repository Dashboard
Do you manage Directory of Open Access Journals? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!