Search Results (16,002)

This work presents the first complete system for automated six degrees of freedom (6DOF) satellite pose estimation from spatially resolved, ground-based, adaptive optics (AO)-corrected imagery, addressing a key challenge in Space Domain Awareness (SDA). The approach mitigates the need for human labeling by directly regressing satellite orientation and position from blurry, noisy, and deeply shadowed imagery. A multi-stage deep neural network pipeline localizes the satellite, predicts pose, and optionally applies temporal filtering. Networks are trained exclusively on fully synthetic imagery generated from a CAD model, yet generalize effectively to real data, bridging the Sim2Real domain gap. On 137 real, human-labeled test images of Seasat, the model achieved a mean rotation error of $5^{°}$ and a mean image-plane translation error of 21 cm. Slant range error was quantitatively evaluated on synthetic data due to unknown real-sensor parameters. Qualitative evaluation of additional real Seasat imagery rated 177 of 199 predicted poses as “ground truth equivalent” or “high-confidence match,” with zero catastrophic failures. The system was extended to seven degrees of freedom (7DOF) for satellites with articulating components and demonstrated on real Hubble Space Telescope (HST) imagery, achieving ${5.5}^{°}$ rotation error, 51 cm image-plane translation error, and $8^{°}$ symmetry-adjusted solar array error on a 249-frame pass with causal temporal filtering. Across 586 real test images from Seasat and HST (captured over multiple decades under diverse conditions) the system consistently performed well. Full 6DOF performance was quantified on a high-fidelity wave optics (HFWO) synthetic test set of Seasat, where the model achieved ${8.4}^{°}$ mean rotation error, 34 cm image-plane translation error, and 1.4% line-of-sight range error at $r_0=6$ cm and 1031 km range. In a limited 200-image benchmark, the model demonstrated 48% lower mean rotation error than a single human labeler while operating ∼800× faster. It required <40 h and a single A100 GPU to generate data and train. The approach was also demonstrated for ARGOS, a smaller satellite with highly symmetric geometry. An exploratory General Image-Quality Equation-based image quality metric (AO-IQ) was introduced as an empirical correlate for pose accuracy. General-purpose models like GPT-4o and Depth Anything V2 failed across most SDA tasks, but rapid gains in vision-language models warrant continued monitoring. These results establish a new operational baseline for practical, real-time satellite pose estimation from AO SDA imagery. Full article

While artificial intelligence (AI) is expected to drive paradigm-shifting transformations, many initiatives result in merely incremental optimization. Anchored in strategic human capital theory, this study shifts the analytical focus from the scale of elite technical talent, conceptualized as AI stars, to the configuration of their knowledge structures to unpack this paradox. Using a dataset of 1270 medical AI patents from corporate R&D teams, we employed high-dimensional fixed-effects models to examine these dynamics. The results reveal that while the knowledge diversity of AI stars acts as a potent engine for breakthrough innovation, this generative capacity is attenuated by excessive domain familiarity. Specifically, direct domain familiarity (derived from internal experience) and indirect domain familiarity (absorbed through external collaborative networks) negatively moderate this relationship, a dynamic theorized to operate through internal cognitive entrenchment and external relational conformity, respectively. Extending the efficiency-driven consensus regarding bilingual expertise, these findings demonstrate that excessive domain embeddedness transforms from an informational bridge into a restrictive constraint during paradigm-shifting innovations, particularly within highly institutionalized environments. Full article

The last decade has seen a rapid increase in individuals turning to artificial intelligence (AI) for advice related to their personal development, especially with the introduction of general-purpose large language models (LLMs) to the general public in 2022. This narrative review examines the potential benefits and risks of using AI for coaching purposes in educational and health contexts. Given that empirical studies on using general-purpose LLMs in these domains remain limited, this paper first synthesizes findings from purpose-built coaching chatbots designed to perform specific tasks that facilitate personal development in these domains and discusses limitations associated with studies that use older versions of chatbots. The reviewed evidence suggests that purpose-built chatbot coaching systems may have some benefits as they are generally well received and can support short-term motivation and selected behavior change, but effects for sustained, meaningful outcomes are inconsistent. We then reflect on the potential risks of using general-purpose LLMs as a coach without appropriate human oversight, by reviewing features of general-purpose LLMs, such as sycophancy, accuracy, and problematic patterns of use. Synthesizing these findings, we consider their implications before identifying potential directions for future research. Full article

Background: Population aging is driving a convergent rise in three disorders historically managed in isolation: cardiovascular disease, neurocognitive decline, and osteoporotic bone loss. Mechanistic data indicate that these systems are coupled through shared regulators of calcium trafficking, inflammation resolution, vascular integrity, and inflammaging. On this basis, a “Heart–Brain–Bone” axis has been proposed; it should be understood as an integrative conceptual framework that organizes evidence drawn from three separate studies, not as a validated physiological entity with agreed diagnostic criteria or demonstrated modifiability. Three lipid-soluble nutrients—long-chain omega-3 polyunsaturated fatty acids (EPA/DHA), vitamin D3 (cholecalciferol), and vitamin K2 (menaquinone-7 [MK-7])—act on overlapping nodes of this network. Methods: We performed a structured narrative review. PubMed/MEDLINE, Embase, the Cochrane Library, and Web of Science were searched from database inception to 25 June 2026 using predefined term blocks for each nutrient, each organ domain, and each candidate mechanism, and the search was updated on 7 August 2026. Records were screened against prespecified inclusion and exclusion criteria by two authors independently, with disagreements resolved by a third. The strength of evidence for each nutrient–organ relationship was graded with an explicitly defined four-level scheme ((−) to (+++)) applied separately to preclinical, observational, randomized and meta-analytic evidence. Results: Vitamin K2-dependent gamma-carboxylation of matrix Gla protein (MGP) and osteocalcin has been proposed to influence whether calcium is incorporated into the bone matrix or deposited in the arterial wall, offering a candidate mechanistic account of the “calcium paradox” associated with isolated vitamin D3 supplementation; EPA/DHA-derived specialized pro-resolving mediators may support resolution of endothelial and neuronal inflammation; and bone-, vascular- and brain-derived signals (osteocalcin, FGF23, the neurovascular unit) interconnect the three organs. These mechanisms are biologically plausible but remain insufficiently confirmed in humans. The clinical evidence is heterogeneous, formulation- and population-dependent, and comprises positive, neutral and null results: cardiovascular omega-3 trials are discordant (REDUCE-IT, which used icosapent ethyl [an EPA ethyl ester], positive; VITAL/STRENGTH/ASCEND null, predominantly in lower-risk or replete cohorts); cognitive trials are largely null or subgroup-dependent (MAPT, DO-HEALTH, VITAL); and MK-7 improves surrogate bone and calcification biomarkers and slowed coronary artery calcification in one recent randomized imaging trial (VitaK-CAC), whereas combined MK-7 plus vitamin D3 did not slow aortic valve or coronary calcification in AVADEC and MK-7 did not reduce bone loss in early menopausal women. Recognized safety signals include a dose-dependent increase in atrial fibrillation with high-dose omega-3, adverse skeletal effects of high-dose or bolus vitamin D, and clinically relevant interference of even low-dose MK-7 with vitamin K antagonist therapy. Conclusions: No adequately powered randomized trial has demonstrated that the combination of long-chain omega-3, vitamin D3 and MK-7 is superior to its individual components or to placebo for any clinical endpoint. The combined regimen is therefore mechanistically rational and hypothesis-generating rather than clinically established; benefit appears most plausible in individuals with elevated risk or demonstrable nutritional insufficiency, and least in replete, low-risk populations. Findings should be interpreted within a broader healthy-aging context that includes lifestyle and psychosocial factors. Adequately powered factorial randomized controlled trials stratified by baseline Omega-3 index, 25(OH)D and vitamin K status, with prespecified mechanistic biomarkers and hard endpoints, are required. Full article

The paper investigates the asymptotic stability for the control of systems described by multi-order fractional differential equations. By utilizing generalized Lyapunov functions and the Kalman–Yakubovich–Popov lemma, frequency-domain criteria are derived to evaluate asymptotic stability. The formulated criteria are similar to the ‘Popov Circle Criterion,’ but the circle parameters are determined by the system’s fractional order and the control parameters. Additionally, the asymptotic stability condition requires that all polar plots associated with the multi-fractional-order system lie inside the circle defining the criterion. Human–Robot System applications highlight the investigation techniques and the particularities of the presented criteria. Full article

Background/Objective: Asthma-related quality of life is related to multiple clinical, physiological, and behavioral factors. Nonetheless, the domain-specific associations of asthma control, inhaler adherence, lung function, and demographic characteristics remain poorly understood. This study aimed to examine the adjusted associations of asthma control, inhaler adherence, lung function, and age with individual domains of asthma-related quality of life. Methods: This cross-sectional observational study included 92 adults with asthma who demonstrated correct inhaler technique and adequate knowledge of their prescribed inhaler regimen. Asthma control was measured using the Asthma Control Test (ACT), inhaler adherence using the Test of Adherence to Inhalers (TAI), and quality of life using the Asthma Quality of Life Questionnaire (AQLQ). Pulmonary function was determined through spirometry. Spearman correlation analyses were initially conducted to identify associations between variables. Subsequently, separate multivariable linear regression models were created for each AQLQ domain, including ACT score, TAI score, age, and FEV₁ % predicted as predictors. Results: A total of 92 adult patients with asthma were studied. Asthma control was significantly positively correlated with all AQLQ domains (all p < 0.001). In multivariable analyses, ACT had the largest absolute standardized coefficient within each of the four domain-specific models (standardized β range: 0.339–0.791). After adjustment, age was associated with the activity limitation and symptom domains, whereas FEV₁ % predicted was associated only with the environmental stimuli domain. Inhaler adherence showed an inverse association with emotional functioning in the fully adjusted model (B = −0.094, 95% CI −0.158–−0.031; p = 0.0038). However, this association was absent in the bivariate analysis and in a sensitivity model excluding ACT (B = −0.015, 95% CI −0.067–0.037; p = 0.562), suggesting a possible suppression effect. Conclusions: ACT showed the largest standardized association within each domain-specific model, although the magnitude of its associations with the symptom and activity-limitation domains may partly reflect conceptual overlap between the instruments. The inverse TAI–emotional-functioning coefficient was model-dependent, did not meet the conservative multiplicity benchmark, and should be considered exploratory and hypothesis-generating. Additionally, lung function and age showed domain-specific associations, highlighting the need for a comprehensive, multidimensional approach in asthma management. Full article

The rapid deployment of low-Earth-orbit mega-constellations has increased the demand for reliable and scalable orbit-anomaly monitoring. Existing methods are vulnerable to heavy-tailed measurement errors, maneuver-induced propagation drift, and the anisotropic uncertainty of short observation arcs. This study proposes an uncertainty-aware Starlink monitoring framework that combines residual-domain wavelet shrinkage with a Huber-weighted adaptive two-body extended Kalman filter to suppress non-Gaussian contamination without obscuring abrupt state changes. A linear altitude correction and a quadratic phase-time correction are introduced into Simplified General Perturbations-4 propagation to compensate for maneuver-related forecast drift. A cross-time covariance model and a joint normalized innovation squared test are further constructed for uncertainty-aware short-arc maneuver sensing, while hierarchical evidence fusion supports anomaly detection and event interpretation. Across paired experiments, the filtering chain reduces position root-mean-square error from 752.4 ± 77.7 m to 175.9 ± 9.9 m, and compensated propagation reduces the 72 h prediction error from 128.7 km to 19.6 km. The detector achieves 93.4% accuracy with a 2.7% false-alarm rate and maintains empirical short-arc false-alarm probabilities near the nominal one percent level. These results demonstrate a consistent engineering link between catalog-scale screening and uncertainty-aware short-arc maneuver sensing. Full article

Large language models (LLMs) now generate fluent natural language and source code, creating challenges for authorship attribution, academic integrity, and software supply-chain security. Most existing detectors for AI-generated content are evaluated separately on natural language or source code, often under matched train–test conditions that can overestimate real-world reliability. We present a paired-prompt benchmark for human-versus-machine detection across English text, Python code, and mixed text–code documents. The benchmark includes $22,141$ instances from HC3, CodeSearchNet, MBPP, and HumanEval across training, validation, and test partitions, plus Mix-Eval, a mixed-content set of 997 Jupyter-notebook-style samples. We evaluate RoBERTa-large for text, GraphCodeBERT and CodeBERT-base for code, a unified RoBERTa-base detector trained on both modalities, and zero-shot baselines. Fine-tuned detectors achieve near-perfect in-distribution performance, with AUROC $1.0000\pm 0.0000$ and accuracy above $99.5\%$. Across five instruction-tuned generator families of varying size (3.8B–7B) and architecture, with the human and problem distributions held fixed, cross-generator transfer causes negligible degradation (AUROC spread $0.0002$; drops of at most $0.0003$). In contrast, domain shift is the main failure mode: on MBPP+HumanEval, GraphCodeBERT drops to $0.85\pm 0.02$ AUROC and CodeBERT-base to $0.67\pm 0.02$. On Mix-Eval, the unified detector outperforms a routed text–code pipeline by 21 AUROC points ($0.96$ vs. $0.75$), largely because of router failures on mixed inputs. Training-time augmentation improves low-false-positive performance, while legacy supervised detectors show systematic class inversion on modern LLM outputs. These results show that reliable deployment requires cross-domain evaluation, mixed-content testing, and calibration beyond in-distribution accuracy. Full article

Quadrotors performing sensing missions near structures such as turbines, towers, and buildings must hold a stable attitude while traversing the structured wind wakes generated by these structures. To maintain trajectory tracking in wind, the vehicle must tilt continuously, and the inner-loop controller must work harder to hold that tilt against the fluctuating flow, raising mean tilt, angular jerk, and command-rate activity. These attitude-domain costs can degrade onboard imagery and gimbal-stabilized sensor data, consuming the actuator authority required to reject further disturbances. Existing work typically treats the two halves of this problem separately: wind is either estimated locally and compensated reactively, or routed around in fields assumed known a priori, and is rarely validated against attitude-domain metrics on hardware. These approaches are most effective when coupled through a single shared representation. A nonlinear disturbance observer estimates wind from the vehicle’s translational dynamics and accumulates it into a spatial map, which simultaneously provides per-stage feedforward compensation to a contouring controller and weights a wind-aware A* planner. On hardware, the estimator matches anemometer ground truth to within $1\mathrm{m}/\mathrm{s}$, and a $2\times 2$ ablation study across three wind configurations shows a reduction of up to $38\%$ in tilt RMS and $28\%$ in its 95th percentile relative to a wind-naive baseline, at the cost of longer paths. Full article

Human action recognition relies heavily on large-scale annotated video datasets, which are costly and time-consuming to curate, while AI-generated videos offer a promising alternative data source, their effectiveness for training action recognition models remains insufficiently explored. This study evaluates AI-generated videos for action recognition across convolutional and transformer-based architectures using real, synthetic, and hybrid datasets. To ensure generative diversity and consistency, a structured prompt engineering pipeline combining action descriptions, environmental contexts, and camera viewpoints was developed. Synthetic datasets were generated using the Grok Imagine and Meta AI Vibes video generation models and paired with a 15-class subset of the Human Motion Database 51 (HMDB51) to construct the Generative Synthetic Human Action Recognition Dataset (GenSynth-HARD). To mitigate domain shift arising from discrepancies between real and AI-generated videos, a Conditional Domain Adversarial Network (CDAN) with a dynamically scaled Gradient Reversal Layer (GRL) was integrated for domain feature alignment. The best-performing hybrid model achieved a Top-1 accuracy of 79.92% on the HMDB51 subset, demonstrating that incorporating synthetic videos effectively supports model performance while significantly reducing annotation overhead. Full article

Objectives: This study aimed to characterize training coverage across different knowledge and skill domains among clinical pharmacists in China and identify variations across domains, providing nationwide evidence to inform training curriculum optimization and future development of clinical pharmacist training programs. Methods: A nationwide questionnaire survey was conducted using a multistage sampling method to collect data on demographic characteristics, training experiences, and self-reported training coverage across 13 knowledge and skill domains among clinical pharmacists. Descriptive statistics summarized the sample characteristics, and subgroup analyses compared differences in training coverage among clinical pharmacists with different demographic characteristics and training experiences. Results: A total of 704 valid questionnaires were included in the statistical analysis. Overall, pharmacists reported high training coverage in clinical pharmacy (97.30%), prescription review (94.60%) and pharmaceutical care (94.18%), whereas training coverage in molecular biology and genomics (53.27%) remained relatively limited. Subgroup analyses indicated significant differences in training coverage across multiple knowledge and skill domains by training level (national vs. provincial, p < 0.05), training type (general vs. specialized, p < 0.05), and certification status (p < 0.05). Conclusions: This study provided the first nationwide characterization of self-reported training coverage among clinical pharmacists in secondary and tertiary hospitals in China across different knowledge and skill domains. The findings provide baseline evidence on the distribution and characteristics of clinical pharmacist training coverage, identify knowledge and skill domains with relatively lower training coverage such as molecular biology and genomics, and inform future optimization of training curricula and training pathways. Full article

High-risk industries (HRI) face persistent safety challenges due to complex environments and multi-factor risks. Traditional manual monitoring is inefficient and reactive. While computer vision (CV) and deep learning (DL) have enabled automated risk perception, existing research lacks systematic review of the transition toward risk cognition. This systematic review, following PRISMA 2020 guidelines, searched Web of Science, Scopus, and IEEE Xplore for studies published from January 2021 to December 2025. After two-stage screening, 108 eligible studies were included. These studies span construction, mining, oil and gas, petrochemical, rail, energy, and maritime industries, with perception-layer applications predominating while cognition and early warning layer studies remain limited. We propose a “perception–cognition–early warning” framework to map the evolution from data to information, knowledge, and actionable decisions. Our findings reveal that visual perception technologies (e.g., Personal Protective Equipment (PPE) detection, object tracking) have matured, but significant bottlenecks persist in multimodal information fusion, knowledge reasoning, and decision-making. Achieving a true cognitive leap requires multimodal semantic alignment, scene graph construction, and causal inference. The proposed framework enables practitioners to design cognitive safety vision systems with scene understanding and interpretable decision-making capabilities. Key implementation strategies include addressing challenges in few-shot learning, cross-domain generalization, and human–machine collaboration. By integrating AI-driven perception, cognitive reasoning, and proactive intervention, this review supports the United Nations Sustainable Development Goals. Relevant goals include Goal 3 (health and well-being), Goal 8 (decent work), and Goal 9 (industry and innovation). Full article

Home-rearing environments are integral to early childhood development. However, it remains unclear whether their developmental significance has changed over time. This study aimed to examine the temporal changes in home-rearing environments and their longitudinal associations with child developmental atypicality in Japan, using baseline assessments from 1999 to 2023 and follow-up assessments through 2024. Data were obtained from the Japanese Childcare Cohort Study, a nationwide childcare-based developmental surveillance project. The analysis included 40,627 1-year observation pairs of 18,210 children enrolled in childcare centers. Home-rearing environments were assessed using the Index of Child Care Environment, and developmental atypicality was evaluated using the Child Development Scale across six domains. Logistic regression models were used to examine the associations between baseline home-rearing environments and developmental atypicality 1 year later, adjusting for baseline developmental status, sex, and age. Generalized additive models and rolling-window analyses were used to assess temporal trends and time-varying associations. Overall, the Index of Child Care Environment scores increased gradually throughout the study period, although social stimulation declined after the mid-2010s. Higher-quality home-rearing environments were generally associated with a lower risk of subsequent developmental atypicality. Protective associations strengthened over time, particularly in the social, communication, vocabulary, and intellectual domains. Home-rearing environments and their developmental implications changed over the 25-year observation period. These findings highlight the continued importance of promoting direct caregiver–child interactions, including shared activities and social engagement. Early childhood surveillance should be periodically re-evaluated to remain relevant to evolving caregiving practices and family contexts. Full article

AI-generated synthetic media are evolving from tools of digital manipulation into a practical antidote for persistent challenges in digital health implementation. However, the computational capabilities that characterise these technologies, their applications and downstream health-related benefits remain fragmented and insufficiently synthesised. This systematic review identified the computational capabilities that characterise AI-generated synthetic media in health, examined their applications and benefits, and developed an integrative framework linking these domains. Twenty-four studies published between 2021 and 31 May 2026 were included. Methodological quality was assessed using the Mixed Methods Appraisal Tool (MMAT) and findings were synthesised through thematic analysis. The synthesis revealed an integrated set of capabilities spanning photorealistic medical-image generation, modality-specific synthesis of clinical images and physiological signals, synthetic non-image health-data creation, preservation of statistical distributions, temporal patterns and clinical relationships, generation of diverse, novel and non-memorised samples and controlled transformation of medical and audiovisual content. Privacy-oriented synthesis and deepfake detection emerged as distinct components supporting privacy-conscious data use, clinical verification and healthcare safety. These demonstrated capabilities were linked to empirically evaluated and indicated applications, including data augmentation, AI model training, diagnostic model development, privacy-oriented health-data sharing, medical education, patient-facing communication, therapeutic support, clinical safety, health-system analytics and planning. The resulting Computational Capability–Application–Benefit (CAB) Framework conceptualises synthetic media as an evidence-graded pathway distinguishing demonstrated computational capabilities, evaluated health-related applications and reported, indicated or potential downstream benefits requiring further validation. AI-generated synthetic media, therefore, represent an emerging computational infrastructure with potential to support safer, privacy-conscious, adaptive and data-intensive healthcare. Full article

This paper presents a load-capacity-constrained arm-angle planning method for a centrally driven humanoid robotic arm. Joint-range and singularity constraints are projected into the one-dimensional arm-angle domain to form a geometric feasible set. A static load-capacity constraint is derived from gravity torque, the Jacobian-transpose mapping of a known endpoint load, and individual actuator-torque limits, and is projected into the same domain. Continuous arm-angle values are then selected along a prescribed Cartesian path within the intersection of the geometric and mechanical feasible sets. In a heavy-load simulation, the maximum output-power metric decreased by 13.3%, and the energy value decreased from 30.60 J to 22.95 J (25.0%). In a prototype proof-of-principle test with a 13 N payload (approximately 1.33 kg), each trajectory was executed three times; the controller-recorded shoulder peak was approximately 1.83% lower, and the recorded energy value decreased from 30.378 J to 28.85 J (5.03%). The prototype values are descriptive because run-level statistics and measurement uncertainty are unavailable, and the experiment covers only one payload and one path. The results support the proposed static, load-capacity-aware planning principle for the tested slow-motion conditions but do not establish general performance or real-time suitability. Full article