Earth and Environmental Sciences

Microalgal-bacterial systems provide a promising strategy to enhance the nitrate removal performance of denitrifiers via the traditionally assumed provision of electron donors or enrichment of functional microorganisms, while the single-species regulatory mechanism remains unclear. Chlorella vulgaris ( C. vulgaris ) could markedly improve the nitrate removal by Paracoccus denitrificans ( P. denitrificans ) even under nutrient-replete conditions, contributing to a 2.48-fold increase in the nitrate removal rate constant compared to the sum of monocultures, indicating non-nutritional regulatory effects. Transcriptomics showed glycolytic genes in P. denitrificans were upregulated by 1.65 to 10.1-fold, accompanied by a 51.0% increase in NADH production. Lactate, as a key intermediate, activated the bacterial electron transport chain (ETC), with ETC-related gene expression upregulated by 2.83 to 2.95-fold. Furthermore, bacterial ammonium induced microalgal glutamate synthesis and release, which likely acted as a signaling molecule to upregulate bacterial nitrate reductase expression. These interactions redirected electrons toward nitrate reduction, increasing carbon utilization efficiency for nitrate removal by 82.75%. Microalgae coordinately optimized bacterial carbon and nitrogen metabolism, reprogramming electron generation, transfer, and allocation to achieve efficient nitrate conversion. This work provides molecular-level mechanistic insights and supports the design of controllable strategies for environmental remediation.

Gas separation is central to industrial processes that drive climate mitigation, clean energy, and sustainable technologies. Metal–organic frameworks (MOFs) offer remarkable tunability for adsorption-based separations, yet identifying optimal materials remains challenging due to their vast structural diversity, costly simulations, and the difficulty of achieving a full range of desired properties. Existing machine learning approaches have accelerated screening but often lack generalizability across diverse gas pairs and operating conditions. Herein, we present BiMix-Bench, a curated database comprising ∼125,900 MOFs and five binary gas mixtures. Leveraging this dataset, LightGBM regressor (LGBMR) models are developed to achieve high predictive accuracy for gas uptakes ( R 2 = 0.93 and 0.92) and selectivity ( R 2 = 0.95) under strict robustness controls, including seed randomness and cross validation. Using CO 2 /H 2 as a case study, we evaluate both zero-shot and few-shot transfer performance. While zero-shot predictions provide limited out-of-distribution accuracy, the pretrained LGBMR models can be efficiently adapted with a small number of new simulations ( N = 204) through transfer learning. This data-efficient adaptation enables the rapid identification of top-performing MOFs, which are subsequently validated through grand canonical Monte Carlo simulations. This generalizable and interpretable framework enables scalable, data-driven discovery of advanced adsorbents for complex and evolving separation tasks.

Urban green space (UGS) users’ perceptions and practices toward biodiversity have undergone significant shifts over the past decade, yet few studies have examined these temporal dynamics. This study analyzes changes in the relationship between UGS users and biodiversity in Paris (2010–2023) using survey data from five green spaces across socially diverse neighborhoods (n = 1217). Our findings reveal a marked evolution in user profiles, with well-being increasingly linked to perceived biodiversity richness. Biodiversity-oriented UGS are now in greater demand, while trust in public policies for biodiversity conservation has declined, accompanied by a growing preference for local community-led initiatives. These results underscore the need for urban planning to account for diverse socio-demographic profiles and their evolving definitions of biodiversity.

Abstract Bumblebees ( Bombus spp.) play an essential role in pollinating a wide range of flowering plants, but many bumblebee species have declined across Europe. Here, we assessed temporal changes in farmland bumblebee community composition alongside increased urbanization of the landscape context over a 19-year period. We repeated a bumblebee survey in 2024 that was originally conducted in 2005 in a farmland area in Helsinki metropolitan area in Finland. We found a significant increase in species richness and a shift in bumblebee community composition over the period. These changes took place in parallel with a significant change in temperature preferences but not tongue lengths or habitat preferences of the bumblebee species. Based on Community Temperature Index (CTI) estimates, the relative abundance of warm-dwelling bumblebee species increased. Most of the new species observed in the study area were species with southern distributions in relation to Finland, suggesting that the increase in species richness may be related to climate change rather than to increased urbanization in the surrounding landscape. Our results suggest that bumblebee community composition may change rapidly with rising temperatures. The increase in species richness and the persistence of several bumblebee species that have declined strongly elsewhere in Europe also suggest that farmland bumblebee species are not necessarily particularly sensitive to urban land use in the surrounding landscape, and suitable habitats in urban areas might play a role in bumblebee species conservation.

Abstract Amid ongoing climate change, the increasing frequency of extreme heat events has become a major concern, particularly in urban areas where the urban heat island effect intensifies thermal risks. Previous studies have largely focused on individual cities, with limited attention to their relationship with landscape patterns. This study investigated the nonlinear relationships between landscape indicators and urban heat risk using machine learning, based on multi-source data from 293 prefecture-level cities in China, and developed a three-dimensional framework of heat hazard, exposure, and vulnerability grounded in the risk framework developed by the Intergovernmental Panel on Climate Change. The results revealed a clear spatial gradient, with higher heat risk concentrated in northwestern China and lower risk in the southeast. Overall, 64% of cities fell into moderate to high-risk categories. Among landscape-related indicators, the building structure index, average height, and average volume AV were key factors associated with heat risk, accounting for 10%, 11%, and 17%, respectively. Partial dependence analysis further indicated nonlinear relationships between landscape patterns and heat risk. This study provides a scalable framework for assessing urban heat risk and offers useful insights for urban planning, particularly in data-scarce regions.

Abstract Blue-Green Infrastructure (BGI) integrates natural elements into urban areas, supporting biodiversity and delivering multiple co-benefits. Its successful implementation, however, requires data-driven ecological planning and coordinated governance, jointly conceived of as social-ecological fit. We assess social-ecological fit using a spatial network framework that combines habitat suitability and connectivity models applied to declining amphibians, with BGI and social network analysis based on multi-sector actor surveys, in Zürich City, a mid-sized European city. Using social-ecological networks (SENs), we identify areas of fit and misfit between BGI and actors. Results show that collaboration among actors is influenced by social-ecological fit, particularly in BGI overlapping with amphibian biodiversity hotspots. However, fit remains limited and uneven: high-value natural and artificial connector habitats, such as natural springs and cemeteries, are often overlooked. The environmental sector serves as a network bridge, while others remain peripheral despite their environmental influence. Our findings highlight the need for cross-sectoral collaboration to enhance biodiversity in cities.

Buildings account for a significant portion of global GHG emissions, yet the building sector lacks high spatial and temporal resolution emissions estimates that could help drive emissions reduction actions. To address this limitation, we propose several methodologies for super-resolving lower-resolution GHG estimates. To examine our proposed disaggregation methods, we utilize the Emissions Database for Global Atmospheric Research (EDGAR) v8.0 gridded emissions data—which exist at 0.1° × 0.1° (approximately 11 × 11 km 2 ) spatial resolution—and subdivide those emissions data into residential and non-residential subsectors to account for critical differences in energy consumption behavior. EDGAR v8.0 gridded data are provided annually from 2015 to 2023, which we spatially super-resolve to a 30″ (approximately 1 × 1 km 2 ) grid and temporally super-resolve using heating degree days to allocate the time-varying portion of emissions at quarterly intervals. To evaluate the accuracy of our proposed disaggregation methods, we use our spatially super-resolved gridded emissions data from EDGAR v8.0 and estimate municipal-level direct onsite CO 2 emissions from buildings across 19,998 municipalities for which emissions data are available. Our spatial super-resolution method provides a two-order-of-magnitude increase in spatial resolution compared to a 0.1° × 0.1° grid cell while decreasing the weighted absolute percentage error from small to large cities as compared to using unmodified EDGAR v8.0 data. These data allow for municipal-level analysis of onsite building emissions for any municipality in the world, revealing that 10% of global direct onsite building emissions are concentrated within only 38 functional urban areas (i.e., a city and the surrounding commuting area), and 25% of these emissions are concentrated within 259 functional urban areas, globally. These municipality-level observations can inform prioritization of subnational emissions reduction actions, particularly in regions that cannot access other forms of emissions inventories.

A train detection system is vital for railway operations and safety, as the braking distance exceeds sight distance. Detection errors can result in signaling deficiencies and potential collisions, such as the near-miss incident at Jiadong station on August 28, 2019. Taiwan Railway employs a fixed-block signaling system with two axle counter subsystems within each block. Discrepancies between these detection systems present challenges for accurate train detection. Various decision logics have been implemented: a serial connection meets fail-safe requirements but reduces efficiency, a parallel connection enhances reliability but compromises safety, and a primary-secondary configuration can lead to underutilized systems. This study proposes an Enhanced Template-Based Voting Logic (ETVL) for dual train detection systems to fully leverage self-diagnosis capabilities and address multiple simultaneous component failures, along with an automatic evaluation platform to assess safety and reliability under these conditions. ETVL employs a Template Construction Module to create comprehensive templates of all possible occupancy statuses by simulating trains passing through the corresponding track, including both normal and failure scenarios. The Occupancy Matching Module then matches real-world occupancy statuses with the preconstructed templates and information from the self-diagnosis system. Case study results demonstrate that ETVL outperforms existing logics in terms of reliability and safety. Implementing this logic can significantly enhance the reliability and safety of dual train detection systems.

Testing hypotheses of phenotypic modularity involves assessing whether groups of traits covary more strongly with each other than with parts outside the group. Structural Equation Modelling (SEM) is a flexible statistical framework for interrogating complex relationships between sets of variables, making it ideally suited to studies of hierarchical modularity and integration. However, quantifying the modular organization of high-dimensional traits using SEM in a phylogenic context has only recently become possible through new methodological advances. Here, we applied SEM to investigate patterns and correlates of phenotypic modularity in the skull and brain of birds. Birds independently evolved relatively large brains multiple times, as well as a wide range of different skull and brain morphologies. While some have proposed the bird skull is composed of several functional or developmental modules, others have suggested the skull is highly integrated, with share allometric scaling structuring trait correlations. The data best supported a model in which brain shape is influenced by changes in shape of the neurocranium as well as a 'jaw' module consisting of the rostrum shape and jaw musculature. Rostrum shape itself does not strongly covary with other aspects of the skull and brain, suggesting decoupling of beak morphology from the rest of the avian cranium. All variables, with the exception of rostrum shape, are strongly influenced by size, supporting the idea that allometry is a major influence on craniofacial integration in birds. These results provide new insights into likely drivers shaping the evolution of the skull in birds and highlight the usefulness of phyloSEM testing hypotheses of evolutionary modularity and integration.

With the increasing concern over greenhouse gas emissions, the penetration of renewable energy sources such as wind, solar, and photovoltaic systems has grown rapidly over the past decade. However, the power output of these renewable units is highly dependent on meteorological conditions, which limits their reliable participation in real-time electricity markets and often reduces the profitability for renewable energy owners. To address the operational uncertainties associated with renewable generation, the concept of a Virtual Power Plant (VPP) has emerged as an effective solution. A VPP aggregates diverse resources including conventional generators, energy storage systems, reserve units, electric vehicles, and flexible demand-side resources to enable coordinated operation and improved market participation. This aggregation allows energy consumption to be shifted strategically, enhancing both consumer benefits and overall social welfare. In this paper, a comprehensive modelling framework for a virtual power plant is developed, integrating stochastic renewable generation with conventional power units. The participating resources operate within predefined confidence bounds and are subjected to reward or penalty mechanisms across multiple time horizons to maintain supply-demand balance while maximizing market profit. The resulting decision-making problem is formulated as a mixed-integer linear programming (MILP) model and solved using a DC load flow approach implemented in MATLAB (version 2015a), employing the Branch and Bound algorithm to handle both continuous and discrete decision variables. The key contribution of this work lies in enabling time-dependent scheduling decisions for VPP components, allowing real-time market settlement through bidirectional communication between the market and generation units. Simulation results demonstrate that such coordinated real-time operation significantly enhances the profit of the VPP facilitator while ensuring system reliability.

Background Muju (睦剧) is a local opera rooted in Chun’an County, Zhejiang Province, China. Since the re-establishment of its professional troupe in 2015, its vocal style has changed substantially, driven by the recruitment of young performers trained in other opera genres (Huangmeixi, Yueju, Wuju). Previous research has described this change through historical narratives and insider accounts, but the extent to which cross-genre training may shape vocal production has remained insufficiently examined with empirical methods. Objective This study examines how performers’ prior vocal training in other operatic genres may affect vocal production when performing Muju, and how this process may have contributed to changes in the genre’s local vocal characteristics over time. Methods A total of 691 vocal excerpts were analyzed, including 196 Muju excerpts across three historical stages and 495 reference excerpts from Yueju, Huangmeixi, and Wuju. A three-layer Transformer-decoder model was used as part of an AI-assisted, score-based acoustic-proximity analysis. The model-derived scores were treated as exploratory indices of relative acoustic proximity rather than as validated measures of stylistic similarity. In addition, a case-based acoustic comparison of the same aria was conducted among three performances: a historical male Muju/Sanjiaoxi reference, a contemporary Huangmeixi-trained female performer, and a contemporary Chun’an local female reference. Results Model-derived scores suggest that Contemporary Muju may show lower relative proximity to Traditional Sanjiaoxi than Old Muju does (Cohen’s d = −0.66), along with stronger Huangmeixi-related score tendencies across historical stages. Performer-level results further suggest variation in Yueju- and Wuju-related model-derived scores by training background. At the case level, the acoustic comparison suggests recurrent differences in vowel openness and articulatory placement among the three performances, especially in the contemporary same-gender contrast between P1 and P7. Conclusion Taken together, the findings are consistent with the possibility that the vocal profile of contemporary Muju has been reconfigured rather than replaced. They are also compatible with the possibility that cross-genre training contributed to this process through the transfer of habitual vocal-production patterns. The acoustic case study provides a focused performance-level illustration, whereas the AI-assisted score-based analysis offers exploratory corpus-level context requiring cautious interpretation.

Objective Insufficient psychological resilience among adolescents has become a critical public health concern that urgently needs to be addressed in China’s basic education sector. Physical education serves as a natural vehicle for implementing frustration education and enhancing adolescents’ psychological resilience. Grounded in Self-Determination Theory (SDT) and indigenous frustration education theory, this study aims to construct a systematic curriculum system integrating situational frustration education into junior high school physical education, and examine its intervention effect on students’ psychological resilience through a 16-week parallel controlled teaching experiment, so as to provide theoretical basis and practical reference for junior high school physical education to achieve the goal of integrated development of physical and mental education. Methods Using a cluster sampling method, a total of 114 students from two parallel Grade 8 classes at No. 2 Middle School of Xuyong County, Luzhou City, Sichuan Province were recruited as participants. One class was randomly assigned to the experimental group ( n = 57) and the other to the control group ( n = 57). The experimental group received physical education integrated with situational frustration education, while the control group received conventional physical education. The Resilience Scale for Chinese Adolescents (RSCA) was administered to both groups before and after the experiment. Independent-samples t-tests, paired-samples t-tests, repeated-measures mixed ANOVA, and analysis of covariance (ANCOVA) were performed using SPSS 29.0 software. Results (1) Repeated-measures mixed ANOVA revealed a significant Time × Group interaction effect on the total psychological resilience score ( F = 6.896, p = 0.010, partial η 2 = 0.058). ANCOVA results showed that after controlling for baseline scores, the post-test total psychological resilience score of the experimental group was significantly higher than that of the control group ( p < 0.001). (2) In the experimental group, students’ scores on the dimensions of Emotion Control, Goal Focus, Interpersonal Assistance, and Positive Cognition improved significantly after the experiment ( p < 0.05), while no significant change was observed in the Family Support dimension ( p > 0.05). (3) In the control group, students’ scores on Emotion Control, Goal Focus, and Positive Cognition improved significantly ( p < 0.05), while no significant changes were found in Interpersonal Assistance and Family Support ( p > 0.05). (4) Situational frustration education significantly improved psychological resilience in both male and female junior high school students, with no significant gender difference in the intervention effect ( p > 0.05). Conclusion Integrating situational frustration education into physical education can significantly improve the overall psychological resilience of junior high school students, and its effect is significantly better than that of conventional physical education. It exerts differential improvement effects on various dimensions of psychological resilience, significantly enhancing students’ abilities in emotion control, goal focus, interpersonal assistance, and positive cognition, making it an effective practical pathway for implementing mental health education objectives in junior high school physical education.

Introduction With the continuous integration of multimodal artificial intelligence (AI) into higher education, students majoring in Early Childhood Education (ECE) are expected not only to master the operational use of multimodal AI tools but also to develop the ability to evaluate, reflect on, and adapt to intelligent learning environments. However, existing studies have mainly focused on engineering and information technology contexts, while limited attention has been paid to ECE students and to the mechanisms through which multimodal AI self-efficacy and cognitive engagement influence academic achievement. Methods This study focused on university students majoring in ECE and collected 458 valid responses through a questionnaire survey. Partial Least Squares Structural Equation Modeling (PLS-SEM) was used to examine the effects of multimodal AI literacy, critical thinking, and resilience on academic achievement, as well as the mediating roles of multimodal AI self-efficacy and cognitive engagement. Results The results showed that multimodal AI literacy, critical thinking, and resilience had significant positive effects on academic achievement. Multimodal AI self-efficacy and cognitive engagement played important mediating roles in the relationships between these antecedent variables and academic achievement. Among the antecedents, critical thinking had the strongest effects on both cognitive engagement and multimodal AI self-efficacy, whereas the direct effect of resilience on cognitive engagement was not significant. Discussion These findings enrich the theoretical understanding of how academic achievement is formed in AI-supported educational contexts. The study also provides practical implications for universities seeking to optimize ECE curricula, enhance students’ multimodal AI application skills, and cultivate their broader competencies for learning and professional development in intelligent educational environments.

Introduction This study aims to examine how individual, organizational, and social resources jointly shape the professional core competencies of college English teachers from a configurational perspective. It integrates Expectancy-Value Theory, the Job Demands-Resources Model, and Social Capital Theory to explain the formation of professional core competencies. Methods A questionnaire on the influencing factors of professional core competencies among college English teachers was developed and validated. Data were collected from 561 college English teachers across 17 universities in East, Central, and Southwest China through stratified random sampling combined with snowball sampling. Fuzzy-set Qualitative Comparative Analysis was employed to explore the configuration effects and pathways of these influencing factors. Results The findings show that external training support and self-efficacy-professional commitment serve as recurrent core conditions across sufficient configurations rather than as single necessary conditions. Other factors demonstrate resource complementarity effects, forming four effective configuration pathways: comprehensive empowerment, collaboration-policy dual-core, policy-efficiency support, and incentive-deficit compensation. These pathways indicate that professional competency formation is characterized by equifinality, resource complementarity, and functional substitution. Discussion The high-level development of professional core competencies among college English teachers results from the synergistic coupling of personal resources, organizational resources, and social capital. The findings suggest that higher education administrators should optimize teacher training systems, improve incentive mechanisms, build collaborative platforms, and provide time management support. Theoretically, this study extends research on teacher professional development by demonstrating the value of configurational causality in understanding professional core competencies.

Introduction This study examined the relationships among early teacher identity, social empathy, and teacher efficacy for inclusive practices among pre-service teachers. Methods A quantitative, cross-sectional survey design was employed. The study sample consisted of 510 undergraduate pre-service teachers (267 early childhood education and 243 primary education students) from seven Turkish universities during the 2025–2026 academic year. Data were collected using the Attitude toward Teaching Profession Scale, the Social Empathy Scale, and the Teacher Efficacy for Inclusive Practices Scale (TEIP). Data analysis was conducted using Partial Least Squares Structural Equation Modeling (PLS-SEM) with SmartPLS 4.0. The analysis followed a two-stage assessment of measurement and structural models and included a 5,000-resample bootstrapping procedure to test the hypothesized relationships. Results Findings revealed that early teacher identity was significantly and positively associated with both social empathy and teacher efficacy for inclusive practices. In addition, social empathy demonstrated a statistically significant indirect effect in the relationship between early teacher identity and teacher efficacy for inclusive practices. The model explained 43.3% of the variance in TEIP. Discussion The results suggest that teacher education programs may benefit from greater emphasis on identity development and socio-emotional competencies, particularly social empathy, in preparing pre-service teachers for inclusive education.

Emotion and cognition analysis in educational interventions is crucial for enhancing personalized learning outcomes. However, existing models often encounter challenges in multimodal data integration and adaptive strategy optimization. To address these challenges, we propose the SymCART model, an intelligent educational intervention framework that integrates multimodal data fusion with deep reinforcement learning-based strategy optimization. SymCART dynamically adjusts teaching strategies through the collaborative operation of a multimodal perception encoder, dynamic cognitive-affective graph inference engine, and adaptive teaching strategy optimizer, thereby improving student learning outcomes. Experimental results demonstrate that SymCART achieves higher predictive accuracy and more effective strategy recommendations compared to traditional models, with statistically significant improvements in AUC, RMSE, weighted F1 for predictive tasks, and nDCG and ADR for policy/recommendation tasks across the IMPROVE, student learning behavior, and additional validation datasets. Ablation studies further confirm the essential contribution of each module, particularly regarding multimodal fusion and strategy optimization. The SymCART model provides robust support for personalized educational interventions and exhibits broad applicability for emotion and cognition analysis as well as adaptive learning strategies.

Objective This study aims to systematically compare the pitch recognition abilities of Mandarin-speaking children with cochlear implants (CI) and their normal-hearing (NH) peers, examining performance variations across different pitch levels (high, normal, low) and Mandarin lexical tones (T1–T4). Methods A comparative experiment was conducted involving 25 children with CI and 25 age-matched NH children. Using acoustic analysis, pitch recognition was assessed through a task where participants identified pitch levels in sentences representing the four Mandarin tones. Results Children with CI demonstrated a dissociated pattern of pitch recognition. Their performance in recognizing high pitch was comparable to that of NH children. However, they exhibited significant deficits in recognizing both normal and low pitch. Furthermore, the type of Mandarin lexical tone did not significantly modulate pitch recognition accuracy in either group. Conclusion The findings indicate a selective deficit in pitch recognition among children with CI, primarily affecting the normal and low pitch ranges. The results provide empirical evidence for developing targeted auditory rehabilitation strategies that address the specific pitch recognition challenges faced by pediatric CI users.

Objective This study aims to examine the relationship between physical education and sports (PES) teacher candidates’ levels of meaning in life and smartphone addiction, and to explore in depth how this relationship occurs through qualitative data. Methods The study was conducted using the explanatory sequential design, one of the mixed-method designs. The quantitative part of the study included 392 PES teacher candidates studying at four universities in Turkey. Quantitative data were collected using the Meaning in Life Questionnaire and the Smartphone Addiction Scale—Short Form. Results The findings showed a low-level, negative, and significant relationship between the presence of meaning and smartphone addiction, whereas no significant relationship was found between the search for meaning and smartphone addiction. Additionally, a low-level, positive, and significant relationship was determined between the presence of meaning and the search for meaning. In the qualitative phase of the research, semi-structured interviews were conducted to support the quantitative findings. Participants’ views indicated that sports provide individuals with meaning in life and partially balance smartphone use. Conclusion As a result, it was determined that the concept of the search for meaning is influenced by the sports context and the sample’s characteristics, and is not a direct risk factor for smartphone addiction; rather, the search for meaning may be an indicator of adaptation or personal development, not merely a lack of a meaningful life. In contrast, it was concluded that experiencing a high level of meaningful life among PES teacher candidates has a protective effect against the risk of smartphone addiction.

Background Functional magnetic resonance imaging neurofeedback (fMRI NF) is gaining popularity as an experimental treatment for psychiatric and neurologic disorders. However, concerns that fMRI NF acts as a placebo rather than through operant conditioning principles, as known for electroencephalography NF, have largely been disregarded. Methods To examine these concerns, 25 healthy individuals underwent training to downregulate activity in the amygdala and subgenual anterior cingulate cortex, and to upregulate both regions concurrently as a bidirectional control condition, across six sessions in a randomized, single-blind, within-subject design. Results While we found no consistent changes in brain activation or training effects across participants, subjective ratings of regulation success mediated the effect of brain activation and objectively achieved regulation on mood changes for both brain regions and regulation directions. Quantified emotional valence and arousal of the regulation strategies had no detectable influence at any stage. Conclusion Our findings suggest predominantly non-specific effects of fMRI NF on mood changes in healthy individuals, which are mediated by the impression of subjective regulation success—at least in the absence of training effects. While the results partially support concerns about the specificity of fMRI NF, further research is required to determine how these findings generalize across different training protocols and patient populations.

Introduction The psychosocial adaptation of rural left-behind children has become a significant social concern. While previous research has mainly focused on developmental vulnerabilities from a deficit perspective, less attention has been paid to the external resources and positive individual traits that facilitate optimal development. To address this gap, grounded in the positive youth development perspective, this study constructed a moderated mediation model to examine the mediating role of stress mindset in the relationship between external assets and the psychosocial adaptation of left-behind children, as well as the moderating role of environmental sensitivity in this process. Methods A questionnaire survey was conducted using a stratified cluster random sampling approach among 2,639 rural children, with 1,590 left-behind children in the target group and 1,049 non-left-behind children in the control group. Data were collected using standardized instruments, including the External Assets Subscale, the Stress Mindset Scale, and the Environmental Sensitivity Scale. Results (1) The levels of external assets and psychosocial adaptation in left-behind children were significantly lower than those of their non-left-behind counterparts; (2) external assets were positively associated with psychosocial adaptation among left-behind children, and stress mindset partially mediated this association; (3) Environmental sensitivity moderated the association between stress mindset and psychosocial adaptation; specifically, this association was stronger among children with higher environmental sensitivity. Conclusion These findings suggest that external assets are positively associated with psychosocial adaptation via stress mindset, with environmental sensitivity potentially moderating this indirect association. Moreover, the positive association between stress mindset and psychosocial adaptation was stronger among children with higher environmental sensitivity.

Background and objective Traditional creativity assessments are limited by subjectivity and high labor costs. Although machine learning (ML) offers objective alternatives, its application to EEG-based creativity evaluation remains scarce. This study aimed to classify high and low creative thinking from EEG signals using ML. Methods One hundred forty participants completed the Alternative Uses Task during EEG recording. Three independent raters (none were authors) evaluated response originality using the Consensus Assessment Technique on a 1-to-5 scale; mean scores were dichotomized at the median into high- and low-creativity labels (996 and 1,096 trials, respectively, from 2,092 valid trials). Classification features included alpha-band Power Spectral Density (PSD), Approximate Entropy, Sample Entropy, and a combined feature set. Six classifiers—Support Vector Machine (SVM), Quadratic Discriminant Analysis (QDA), Logistic Regression (LogR), Decision Tree (DT), XGBoost, and LightGBM—were trained and evaluated using a 10-fold cross-validation strategy. To prevent subject-level information leakage, a Leave-One-Subject-Out (LOSO) validation was additionally conducted. Results All six classifiers effectively distinguished creativity levels. Under 10-fold cross-validation, SVM achieved optimal performance using Approximate Entropy or Sample Entropy ( F 1-score = 90.5%; accuracy = 89.8%). The combined feature set yielded comparable results. LOSO validation confirmed generalizability to unseen individuals, with SVM attaining F 1-scores of 82.4% (Approximate Entropy) and 82.1% (Sample Entropy). Entropy-based features consistently outperformed alpha PSD. Conclusion ML effectively classifies creativity from EEG signals. The superior performance of entropy features, supported by both trial-level and subject-independent validation, highlights the robustness of the proposed approach and its potential for developing objective, scalable creativity assessment tools.

The integration of Generative AI (GenAI) into English as a Foreign Language (EFL) teaching is rapidly expanding, yet its effects on learners' affective states—key to language acquisition from a positive psychology lens—remain inadequately summarized. This systematic review synthesizes empirical evidence on how GenAI tools influence motivation, anxiety, and well-being among EFL learners in higher education. Following PRISMA guidelines, 1,420 records were identified from four databases (2019–2025), with 29 studies meeting inclusion criteria after screening. Results reveal a generally positive affective impact: GenAI enhances motivation and engagement, while typically reducing anxiety—though one study noted increased speaking anxiety. Benefits also extend to well-being, including emotional support and mindfulness. However, effects are moderated by factors such as gender (higher affective scores and AI self-efficacy in males) and academic level (lower perceived utility among Master's students). The findings affirm that well-designed GenAI aligns with Self-Determination Theory by supporting autonomy, competence, and relatedness. In conclusion, GenAI shows strong potential to foster positive psychological outcomes in EFL learning, yet its implementation must be intentionally tailored and inclusively designed to address contextual and individual differences.