Search Results (2,879)

Intrinsically self-sensing cementitious composites offer a promising basis for continuous structural health monitoring. Their electrical response, however, is strongly affected by reversible moisture change and freeze–thaw exposure. This study presents a computational proof-of-concept for the AI-driven design of a federated digital-twin architecture for damage identification and adaptive sensing in sustainable self-sensing cementitious infrastructure. The framework is developed and evaluated entirely in software on a physics-based synthetic testbed. At its present maturity, it is therefore a digital-twin precursor rather than an operational digital twin: it has no calibrated physical counterpart and no live, two-way data coupling, and no experimental validation is claimed. A transparent, physics-based signal generator produces fractional-change-in-resistance signals for twelve virtual CNT/biochar-functionalized LC3 and geopolymer specimens. Each passes through four progressive damage stages interleaved with wet–dry and freeze–thaw conditioning. The framework integrates a CNN-LSTM damage classifier, unsupervised domain adaptation, federated learning, reinforcement-learning-based active sensing, and quantum-inspired aggregation optimization. On three unseen virtual specimens (654 evaluation windows), the CNN-LSTM achieved 70.3% four-stage accuracy (95% Wilson confidence interval 66.7–73.7%) and a macro-F1 score of 0.650, with per-specimen accuracy ranging from 63.8% to 77.1%. It reached 85.2% (95% CI 82.3–87.7%) for the damaged-versus-undamaged decision and reduced environment-induced false alarms by 74.7% (95% CI 61.7–83.4%) relative to a calibrated threshold detector. Federated averaging was less accurate and less stable than centralized training; the 5.2 percentage-point gain from quantum-inspired aggregation lies within the resolution of the evaluation set and is not established as a real improvement. The active-sensing controller reduced measurement cost by 98.9% but detected only four of 27 damage-progression events. All sensing data are synthetic, and every interval reported here is recomputed from the evaluation counts already reported rather than obtained from additional experiments. The results therefore establish algorithmic feasibility only and identify the components requiring refinement before experimental validation. Full article

This article examines how drama in English portrays men who anchor their self-worth to sexual potency and masculine performance, particularly when that confidence faces direct challenge or the quiet pressure of ageing. Through sustained close reading of Macbeth, Antony and Cleopatra, and Who’s Afraid of Virginia Woolf?, the analysis traces how characters negotiate pride, vulnerability, and the defence of manhood through dialogue, pacing, and silence. Existing criticism has often approached these moments through historical or ideological lenses; this reading instead attends to the psychological negotiation visible in the text itself. Drawing on established work on gender performance and narrative identity, the article treats masculine pride not as a fixed cultural pattern but as a dramatic act that characters repeat, defend, and sometimes surrender. The theoretical tools serve the text rather than dictate it, allowing the plays to reveal how men construct, protect, and adjust their sense of virility when it is called into question. The findings emerge from careful attention to scene structure, verbal exchange, and the unspoken tensions that shape dramatic portrayal. Full article

Background: Resilience is an important resource in nursing education, supporting students in adapting to demanding academic and clinical environments. However, evidence regarding the factors associated with resilience among nursing students remains inconsistent. This study aimed to examine the factors associated with resilience among undergraduate nursing students, with particular emphasis on gender, age, academic progression, institutional context, and selected sociodemographic characteristics. Methods: A cross-sectional quantitative study was conducted with 263 undergraduate nursing students from two higher education institutions in Portugal. Data were collected using the Takviriyanun Resilience Factors Scale, adapted and validated for the Portuguese population. Descriptive and inferential statistical analyses were performed to examine the two dimensions of the scale—Personal and Social Support Skills and Problem-Solving Skills—and their associations with sociodemographic and academic variables. Results: Students reported generally favourable levels of resilience-related factors, with higher scores observed in items reflecting personal responsibility and perceived social support. Comparatively lower scores were observed for items related to calmness and patience, confidence, optimism, and perceived social acceptance. A statistically significant gender difference was found in Problem-Solving Skills, with male students showing higher mean ranks than female students (p = 0.038); however, the effect size was small (r = 0.13). A moderate positive correlation was observed between Problem-Solving Skills and Personal and Social Support Skills (ρ = 0.394, p < 0.001). No statistically significant associations were found between the resilience dimensions and age, year of study, higher education institution, displacement status, place of residence, or religious affiliation. Conclusions: Undergraduate nursing students demonstrated generally favourable levels of resilience-related factors. Personal and Social Support Skills and Problem-Solving Skills were positively associated, suggesting a meaningful relationship between interpersonal resources and problem-solving competencies. Although a small gender difference was observed in Problem-Solving Skills, the findings do not indicate substantial gender-related differences in resilience-related factors. Educational strategies aimed at strengthening problem-solving, adaptive coping, emotional self-regulation, and social support may be relevant throughout undergraduate nursing education. Longitudinal and multicentre studies are warranted to further investigate the factors associated with resilience and to evaluate the effectiveness of resilience-focused educational interventions. Full article

Background: Handgrip strength has been proposed as an objective marker of post-COVID functional impairment, but its contribution relative to fatigue and sleep quality has not been formally evaluated in Latin American populations. Methods: We conducted an exploratory cross-sectional analysis of 130 adults with post-COVID condition in Palmira, Colombia. The outcome was any functional limitation on the Post-COVID Functional Status (PCFS) scale (grades 1–4 versus grade 0). Eight predictors were pre-specified on clinical grounds before examining outcome associations, and ridge-penalized logistic regression was pre-specified as the primary model. Sample size adequacy followed the criteria of Riley et al. Internal validation used bootstrap optimism correction (B = 1000), re-running the complete pipeline within each replicate. We assessed calibration, decision curves, prediction stability, SHapley Additive exPlanations (SHAP) with bootstrap rank intervals, and subgroup performance. Reporting followed TRIPOD + AI; risk of bias was self-assessed with PROBAST + AI. Results: Functional limitation affected 91/130 participants (70.0%). The minimum sample required for eight parameters was 566; the cohort met 23.0% of this requirement, so the analysis is exploratory. The optimism-corrected area under the curve was 0.748 (95% confidence interval (CI) 0.669–0.830), with calibration slope 1.15 (0.71–1.69). Fatigue ranked first in SHAP importance (54.0% of resamples), whereas grip strength ranked fifth (median rank 6). Net benefit exceeded both default strategies between thresholds 0.38 and 0.88. Thirty percent of participants changed classification in over 20% of resamples. Conclusions: Fatigue, sleep quality and acute-phase severity, rather than grip strength, emerged as the most influential correlates. These hypothesis-generating findings require confirmation in adequately powered cohorts. Full article

Background: Screen-based distractions during meals are increasingly common during adolescence and have been linked to less healthy eating behaviors. Chronotype, an individual’s circadian preference for morningness or eveningness, is associated with differences in alertness, self-regulation, and recreational screen use across the day, which may in turn influence susceptibility to distractions during meals. However, the relationship between chronotype and eating distraction behaviors remains largely unexplored. Therefore, this study examined the association between chronotype and eating distraction behaviors in Spanish adolescents. Methods: A cross-sectional study was carried out in 820 adolescents aged 12–17 years from the Eating Healthy and Daily Life Activities (EHDLA) project. Chronotype was assessed using the Morningness–Eveningness Scale for Children and classified as evening, intermediate, or morning type. Eating distractions were evaluated using a composite score that included eating lunch or dinner while watching television, using mobile devices or social networks during meals, and eating while standing. Associations were examined using ordinal regression models (proportional odds), with eveningness as the reference category; both unadjusted and adjusted models are reported, with adjustment for relevant sociodemographic and lifestyle covariates. Results: In adjusted ordinal regression models with eveningness as reference, morningness was related to lower odds of being in a higher eating distractions category (odds ratio [OR] = 0.53; 95% confidence interval [CI] 0.35–0.82; p = 0.004). Morningness was also associated with lower odds of a higher category of television viewing during meals (OR = 0.59; 95% CI 0.38–0.93; p = 0.024). No significant associations were identified for eating while using a phone or social networks or for eating while standing. Conclusions: Chronotype was independently associated with eating distraction behaviors in Spanish adolescents, with evening types showing a less favorable pattern. These findings suggest that circadian preference may be relevant when addressing mealtime behaviors during adolescence and could help inform future health promotion strategies. Full article

Accurate detection of tomato flowering stages is important for greenhouse crop management and automated pollination, but it remains challenging because tomato flowers are small, densely distributed, frequently occluded, and visually similar across adjacent developmental stages. To address these problems, this study proposes YOLO11-FAL, an improved object detection model based on YOLO11 for tomato flowering stage detection in greenhouse environments. The original C3k2 modules are replaced with C3k2_Faster to reduce redundant spatial computation and enhance local structural feature representation. An Attentional Scale Sequence Fusion (ASF) structure is introduced into the neck network to strengthen multi-scale feature interaction, and a Localization Quality Estimation Head (LQEHead) is incorporated to recalibrate classification confidence using bounding-box distribution information. Experiments were conducted on a self-constructed tomato flower dataset containing Bud, Anthesis, and Post-anthesis stages under varying illumination conditions. YOLO11-FAL achieved 92.00% Precision, 88.88% mAP@0.5, 62.21% mAP@0.75, and 56.72% mAP@0.5:0.95. The model contained 2.384 M parameters and achieved 49.58 FPS on an NVIDIA Jetson AGX Orin (NVIDIA Corporation, Santa Clara, CA, USA) using TensorRT FP16, with a measured onboard power consumption of 11.07 W. These results indicate that YOLO11-FAL provides a practical and efficient visual perception approach for greenhouse tomato flowering stage detection. Full article

Background: Health security among structurally vulnerable populations depends not only on healthcare access but also on livelihood stability, caregiving capacity, and social participation. Evidence on how community-engaged participatory interventions address social determinants of health in low- and middle-income settings remains limited, particularly regarding how such interventions are implemented, documented, and evaluated in practice. Methods: This qualitative implementation study examined a participatory livelihood and caregiving intervention in a rural Thai setting. Over a nine-month period, 50 core participants from structurally vulnerable households engaged in integrated livelihood and caregiving activities. Data were collected through field observations, participatory reflection meetings, implementation records, LINE-group communication archives, and voluntary household financial logs, and were analysed using inductive thematic analysis. Results: Findings revealed perceived improvements across four interconnected dimensions of well-being: caregiving capacity, social connectedness, livelihood flexibility, and household economic buffering. Participants more consistently described household stabilization achieved through reduced expenditure, resource sharing, and livelihood diversification than rapid income growth. Sustained group participation was associated with strengthened peer support networks and greater self-reported confidence in managing caregiving and household challenges; however, these findings rest on self-reported and predominantly qualitative evidence, with financial documentation available for only a small, non-random subset of participants. Conclusions: These exploratory findings suggest that community-engaged livelihood and caregiving interventions may strengthen multidimensional health security by supporting stabilization processes rather than delivering immediate economic gains. Relational support, adaptive implementation, and community participation emerged as candidate mechanisms linking policy intentions to community health responsiveness. Health policy frameworks in comparable low- and middle-income settings may benefit from incorporating structural and relational social determinants alongside income-based metrics, although confirmatory research with comparison groups is needed before these implementation-level insights can inform generalizable policy guidance. Full article

Existing LiDAR super-resolution methods primarily aim to increase point cloud density or improve coordinate reconstruction accuracy. However, they tend to introduce blurred edges and distorted planar surfaces during reconstruction, making it difficult to preserve the consistency of local scene geometry. To address this issue, this paper proposes a structural consistency-aware LiDAR super-resolution method that aims to preserve the local geometric relationships of the reconstructed point cloud with respect to the ground-truth point cloud in edge and planar regions. Specifically, complementary observations from adjacent frames are first fused using multi-scale dilated convolutions. An anisotropic Swin Transformer and a Coordinate-Aware Structure Enhancement (CASE) module are then employed to accommodate the horizontally dense and vertically sparse sampling pattern of LiDAR, strengthen long-range geometric modeling, and reduce the loss of critical structural information. During training, a local curvature-based structural consistency loss is designed to separately constrain edge sharpness and planar smoothness. During inference, prediction uncertainty and point cloud height are combined to adaptively remove low-confidence points, further improving the geometric reliability of the reconstructed point cloud. Experiments on the KITTI dataset show that the proposed method achieves an MAE of 0.4916 and an IoU of 0.4633, outperforming the representative comparison methods on both metrics. When the reconstructed point clouds are applied to A-LOAM, the average RTE and RRE values are reduced by 34.6% and 31.2%, respectively. In addition, experiments on the self-collected CSU-SLAM dataset provide preliminary evidence of the applicability of the proposed method to indoor and outdoor scenes under a different LiDAR configuration. Full article

Pixel-level annotation of remote sensing imagery is costly, motivating weakly supervised semantic segmentation (WSSS) using only image-level labels. However, class activation maps (CAMs) often highlight only discriminative sub-regions and fail to separate adjacent land-cover regions, particularly in remote sensing scenes characterized by densely co-occurring land-cover classes and substantial variations in object scale. To address these limitations, we propose a three-stage framework that integrates complementary priors from Contrastive Language–Image Pre-training (CLIP), Self-Distillation with No Labels version 2 (DINOv2), and the Segment Anything Model (SAM). First, a lightweight CLIP adapter aligns vision–language priors with remote sensing imagery, while sigmoid-based multi-label decoupled distillation replaces class-competitive distillation with independent class-wise supervision, producing more complete CAMs. Second, DINOv2-guided feature clustering decomposes large merged regions before SAM prompt generation, while Spatial–Semantic Constraints are used to construct confidence-guided point-and-box prompts and reject excessively expanded or semantically inconsistent masks, thereby generating reliable pseudo-labels. Finally, a compact segmentation network is initialized with the weights learned in Stage 1 and retrained using the refined pseudo-labels generated in Stage 2, eliminating the need for foundation models during inference. Experiments on the Potsdam, LoveDA, and DeepGlobe datasets show that the proposed method achieves mean intersection over union (mIoU) scores of 53.16%, 52.66%, and 62.98%, respectively, outperforming state-of-the-art WSSS baselines by 6.55, 1.16, and 1.27 percentage points, respectively. These results demonstrate the effectiveness and generalizability of the proposed framework across diverse remote sensing scenarios under image-level supervision. Full article

Spatial literacy and spatial thinking are fundamental skills for understanding space, interpreting geographical phenomena, and enabling active citizenship in a society increasingly driven by geospatial information. This review examines how digital tools and geospatial technologies contribute to cultivating spatial literacy and thinking within primary geography education. Following an integrative, PRISMA-informed review methodology combining searches of Scopus, Web of Science, and ERIC (n = 1373 records identified), this study draws on a corpus of 141 studies—comprising an original set of sources supplemented by newly identified, priority-ranked literature—examining digital maps, Geographic Information Systems (GIS), Web GIS, StoryMaps, collaborative mapping, digital games, and augmented and virtual reality in geography teaching. This review proposes an explicit conceptual model clarifying the hierarchical and complementary relationships among spatial skills, spatial thinking, critical spatial thinking, spatial literacy, and spatial citizenship, together with a four-dimensional typology distinguishing technological platforms, representational formats, pedagogical approaches, and learning activity types; both are proposed as synthesizing heuristics for organizing the reviewed evidence and are offered as a basis for future empirical validation rather than as established models. The findings suggest that pedagogically grounded digital integration strengthens spatial skills, shifts learning toward active, participatory models, and helps cultivate spatial citizenship, though the strength of this evidence varies considerably by tool and by educational level, with primary-specific support remaining comparatively limited for higher-order outcomes such as critical spatial thinking. Crucially, the effective integration of these technologies depends on structured teacher training—addressing self-efficacy and confidence-related barriers alongside technical skills—and the adoption of innovative instructional approaches. Ultimately, digital tools provide effective means of deepening geographical learning and strengthening spatial literacy in primary schools. Realizing this potential, however, depends on matching the selection and use of each tool to the specific dimension of spatial learning it is evidenced to support. Full article

Background: Generative Artificial Intelligence (GenAI) is increasingly integrated into nursing education, yet structured AI literacy training and ethical guidance remain limited. Consequently, nursing students often rely on informal learning, resulting in variability in AI readiness, confidence, and responsible use. Aims: This study was conducted to examine (1) whether AI literacy was positively associated with AI self-efficacy and AI attitudes and (2) whether AI self-efficacy mediated the relationship between AI literacy and AI attitudes. Methods: A cross-sectional survey using convenience sampling was conducted with 100 prelicensure nursing students in New York City. Data were collected using the AI Literacy Scale (AILS), AI Self-Efficacy Scale (AISES), and Generative AI Attitude Scale (GAIAS). Correlation and path analyses were performed using SPSS and Amos 30.0. Results: The participants had a mean age of 30.25 years, and 71% were women. AI literacy and AI self-efficacy were both positively associated with AI attitudes (all p < 0.001). Path analysis showed that AI literacy significantly predicted AI self-efficacy (β = 0.39, p < 0.001) and AI attitudes (β = 0.28, p = 0.003). AI self-efficacy significantly predicted AI attitudes (β = 0.31, p = 0.001) and partially mediated the relationship between AI literacy and AI attitudes. Conclusions: AI self-efficacy partially mediated the relationship between AI literacy and AI attitudes. Nursing curricula may benefit from structured AI education that integrates guided GenAI practice, case-based learning, and faculty feedback. Such educational frameworks warrant further empirical investigation regarding their potential to foster AI literacy, AI self-efficacy, and positive attitudes toward responsible AI integration, particularly through longitudinal studies assessing subsequent behavioral outcomes. Full article

Psychological distress is associated with problematic online shopping, but the relative and independent associations of depression, anxiety, and stress require careful evaluation without causal interpretation. This study examined their associations with continuous Online Shopping Addiction Scale (OSAS) scores after accounting for sociodemographic characteristics and online shopping frequency. A cross-sectional convenience sample included 438 adults in Türkiye (54.0% female among respondents reporting gender; mean age = 33.79, SD = 10.24); a total of 317 participated online and 121 completed printed questionnaires face to face. Pearson correlations were estimated in the full sample. Hierarchical regressions used a common complete-case sample (n = 375). Because residual diagnostics indicated heteroscedasticity and tail departures, HC3 robust standard errors and 95% confidence intervals were reported. Anxiety, depression, and stress showed statistically significant but modest correlations with OSAS total scores (r = 0.34, 0.32, and 0.27, respectively; all p < 0.001). In separate adjusted models, anxiety (ΔR² = 0.070; β = 0.283), depression (ΔR² = 0.068; β = 0.276), and stress (ΔR² = 0.042; β = 0.217) each increased explained variance (all p < 0.001). In the combined model, the DASS-21 block added ΔR² = 0.091; anxiety (β = 0.206, p = 0.002) and depression (β = 0.200, p = 0.001) had comparable independent associations, whereas stress did not (β = −0.050, p = 0.473). Online shopping frequency had the largest standardized coefficient among concurrent covariates (β = 0.280), but it is conceptually proximal to the OSAS outcome. The combined model explained 31.6% of the variance (adjusted R² = 0.291). Recruitment-mode comparisons were small and nonsignificant after Holm correction. Depression and anxiety were independently associated with higher OSAS scores, while most outcome variance remained unexplained. The findings do not establish temporal prediction, causation, or clinical diagnosis and should be interpreted in light of self-report measurement, construct overlap, residual confounding, and the digitally engaged convenience sample. Full article

Innovation is an everyday imperative in the information and communication technology (ICT) sector, yet the psychological conditions through which well-being is associated with innovative action remain insufficiently understood. This study investigates whether subjective vitality and creative self-efficacy constitute distinct psychological pathways—an energy-based and a belief-based mechanism, respectively—through which psychological well-being becomes associated with innovative work behavior, thereby providing a more differentiated examination of the assumption that psychologically healthy employees are more likely to behave innovatively. An online survey was administered to 463 ICT-sector employees in Türkiye, and partial least squares structural equation modeling (PLS-SEM) was used to test direct and indirect relationships with comprehensive assessment of measurement quality, predictive relevance, and effect sizes. Creative self-efficacy emerged as a meaningful indirect pathway linking psychological well-being to innovative work behavior, whereas subjective vitality, though strongly predicted by well-being, showed no significant relationship with innovative outcomes. These findings suggest that not all positive psychological states are equally innovation-relevant: among ICT employees, the indirect association between well-being and innovation was evident through employees’ confidence in their own creative capability rather than through feelings of psychological energy—a distinction with meaningful implications for both theory and practice in knowledge-intensive, digitally dynamic organizations. Full article

Background: The discordance between individuals’ self-perceived risk of HIV infection and laboratory-confirmed risk has a significant impact on behaviours related to HIV prevention and timely initiation of treatment. Methods: Data was obtained from the sixth South African National HIV Prevalence, Incidence, and Behaviour Survey. Descriptive statistics were used to summarise the sample characteristics and provide the prevalence of discordance. A multivariable multinomial logistic regression was employed to investigate factors associated with discordant results. Results: Of 34,528 participants, 4.1% (95% Confidence Interval (CI): 3.7–4.5) were classified as HIV-positive with low perceived risk, while 11.5% (95% CI: 10.4–12.6) were classified as HIV-negative with high perceived risk and 1.0% (95% CI: 0.8–1.1) as self-misdiagnosed. Being HIV-positive with low perceived risk was significant among older ages (25–54 years) and lower among other races than Black Africans and those residing in rural informal areas. In contrast, being HIV-negative with high perceived risk of HIV risk was more common among participants aged 25–44 years, as well as those living in rural informal and rural formal areas, but lower among other races than Black Africans. The factor associated with self-misdiagnosis was being female. However, participants aged 15–24 years who were never married were less likely to self-misdiagnose. Conclusion: These findings highlight a significant discrepancy between perceived and laboratory-confirmed HIV risk, underscoring the importance of targeted interventions to improve risk perception accuracy. Enhancing individuals’ understanding of their true risk may lead to better engagement in preventive measures and the timely initiation of treatment. Full article

This pilot in vitro study explored the tensile retention of lithium disilicate and monolithic zirconia crowns cemented onto zirconia one-piece implants using two resin cements and two cementation protocols. In addition, the relationship between micro-computed tomography (micro-CT)-derived cement layer characteristics and retention was explored for lithium disilicate crowns. Thirty-two implant–crown assemblies were prepared using 16 lithium disilicate and 16 zirconia crowns. Specimens were cemented with either an adhesive resin cement (Panavia V5) or a self-adhesive resin cement (SpeedCem Plus) using two protocols: conventional apical-half cementation (AH) and an abutment-assisted apical-half protocol (A-AH). Cement thickness and porosity for lithium disilicate crowns were obtained from a previously published micro-CT analysis of the same specimens; no micro-CT measurements were available for the zirconia specimens. Tensile pull-out testing was performed using a universal testing machine. The primary outcome was the maximum recorded force at the first observed mechanical failure, irrespective of the mode of that failure, so that all 32 specimens contributed a value. Failure occurred by crown debonding in 27 specimens, by crown fracture in 4 and by implant fracture in 1. For the primary outcome, the maximum recorded force was lower for lithium disilicate than for zirconia crowns (medians 347.20 versus 596.05 N; exact Mann–Whitney p = 0.017) and lower with the A-AH than with the AH protocol (medians 304.24 versus 614.38 N; p < 0.001), whereas the difference between the two resin cements was not statistically significant (medians 438.88 versus 550.83 N; p = 0.210). The highest observed mean maximum load was recorded for zirconia crowns cemented with Panavia V5 using the AH protocol (729.9 ± 237.7 N), whereas the lowest observed mean maximum load was recorded for lithium disilicate crowns cemented with Panavia V5 using the A-AH protocol (219.7 ± 105.1 N). In a secondary, cause-specific exploratory analysis restricted to crown debonding (27 events, 5 specimens censored at fracture), Cox proportional hazards regression on the applied-force scale gave hazard ratios of 3.75 (95% CI 1.40–10.01) for lithium disilicate versus zirconia, 6.47 (2.39–17.53) for A-AH versus AH and 1.82 (0.76–4.39) for Panavia V5 versus SpeedCem Plus. For lithium disilicate crowns, exploratory factorial ANOVA indicated that cementation protocol was associated with differences in cement thickness (p = 0.035), while cement type was associated with differences in porosity (p < 0.001). All 16 lithium disilicate cement thickness observations lay between 253.29 and 254.96 µm, a total span of 1.67 µm. Within that extremely restricted range, a univariable exploratory Cox model expressed per 0.1 µm gave a hazard ratio for debonding of 1.24 (95% CI 1.03–1.48; p = 0.024); this is an unadjusted association across a range that is itself associated with cementation protocol, and it does not demonstrate a clinically meaningful or independent effect of cement thickness. No association was detected for total porosity (0.959 per percentage point, 0.717–1.283); that interval is wide and indicates absence of evidence rather than evidence of no association. Within the limitations of this pilot in vitro study—four specimens per subgroup, wide confidence intervals and no adjustment for multiplicity—the findings suggest that crown material and cementation protocol may be associated with retention patterns. They are exploratory and hypothesis-generating and require confirmation in larger, independently powered studies. Full article