Search Results (18,136)

Industrial large models (ILMs) are reshaping manufacturing toward adaptive, learning-enabled systems. Yet whether ILM adoption confers systemic productivity advantages over conventional digitization remains unexplored from a systems perspective. We develop a socio-technical systems framework to examine differential associations of ILM adoption versus generic digitalization with manufacturing firms’ New Quality Productive Forces (NQPFs), a composite measure of production system transformation toward higher efficiency and sustainability. Analyzing 3847 Chinese manufacturing firms (2012–2025) using a BERT-based NLP measure, we find that ILM adoption is associated with more than double the estimated productivity associations of conventional digitization (0.187 vs. 0.092 SD). These associations operate through three subsystems, including innovation (31.0%), operations (24.6%), and quality assurance (20.3%), and are amplified by skilled labor, industrial software, and technology-intensive settings. Our findings provide systems-level evidence that ILMs are associated with a deeper form of system transformation toward greater productivity and sustainability, rather than merely optimizing existing processes. The empirical analysis focuses on the economic sustainability dimension, as captured by the NQPF index; broader sustainability implications are discussed as theoretical and policy inferences rather than directly measured environmental outcomes. Full article

The convergence of additive manufacturing (AM) and artificial intelligence (AI) is poised to redefine the landscape of modern production; however, the literature remains fragmented across isolated applications, lacking a unified perspective on the engineering impact and practical deployment of these technologies. This review provides a comprehensive and critical synthesis of the state of the art in AI-enhanced AM, systematically covering supervised, unsupervised, and reinforcement learning paradigms, alongside deep-learning-based computer vision, natural language processing, and robotics. In contrast to prior works that focus on singular aspects, this paper consolidates progress across four core engineering domains: (i) lightweight and manufacturable design, (ii) real-time in situ defect detection and process analysis, (iii) energy-efficient process optimization, and (iv) cost-effective build-time estimation with intelligent support minimization. Beyond cataloging these advances, this review identifies key quantitative benchmarks and recurring technical challenges, including data scarcity, poor model generalizability, and the critical gap between offline prediction and real-time closed-loop control. To transcend these isolated successes and enable industrial adoption, we propose a novel, unified closed-loop AI-AM framework that tightly integrates generative design, process planning, in situ production monitoring, and continuous model updating into a cohesive digital thread. Furthermore, a domain-stratified SWOT analysis is compiled, offering a strategic evaluation of strengths, weaknesses, opportunities, and threats across the four application pillars. By bridging the gap between laboratory prototypes and production-ready autonomous systems, this review serves as a definitive reference for researchers and practitioners aiming to navigate, deploy, and advance the rapidly evolving field of AI in additive manufacturing. Full article

Epigenetic regulation has transformed our understanding of how fish adapt to changing environments by modulating gene expression without altering the underlying DNA sequence. This review explores the “dark mastery” of fish epigenetics by providing mechanistic insights into the principal epigenetic processes, including DNA methylation, histone modifications, chromatin remodeling, and non-coding RNAs, that govern development, immunity, stress responses, and disease susceptibility. These regulatory mechanisms enable fish to respond dynamically to environmental stressors such as temperature fluctuations, salinity shifts, hypoxia, pollutants, ultraviolet radiation, and nutritional changes, thereby influencing physiological resilience, reproductive performance, and survival. Recent advances in next-generation sequencing and multi-omics technologies have substantially expanded our understanding of the fish epigenome, while bioinformatics has become indispensable for integrating and interpreting complex genomic, transcriptomic, and epigenomic datasets. Furthermore, artificial intelligence (AI) and machine learning (ML) are emerging as powerful approaches for biomarker discovery, predictive modeling of disease susceptibility, environmental risk assessment, and precision aquaculture. The integration of epigenetics with bioinformatics and AI provides unprecedented opportunities to decipher complex regulatory networks, identify adaptive epigenetic signatures, and develop data-driven strategies for improving fish health and aquaculture sustainability. Despite these advances, important challenges remain, including limited species-specific epigenomic resources, difficulties in multi-omics integration, model interpretability, and the need for standardized analytical frameworks. This review highlights current knowledge, emerging computational approaches, and future perspectives for translating epigenetic discoveries into sustainable aquaculture practices and aquatic ecosystem conservation under accelerating environmental change. Full article

Food quality and safety have become increasingly critical public health concerns, while traditional detection approaches are constrained by laborious sample preparation, lengthy analysis times, and destructive procedures. Hyperspectral imaging (HSI) has emerged as a rapid, non-destructive method that synergistically fuses spatial and spectral information for comprehensive food quality and safety assessment. However, the high dimensionality and nonlinear features of HSI data pose substantial challenges for traditional chemometric models. This review provides a comprehensive overview of recent advances in machine learning (ML) and deep learning (DL) algorithm-assisted HSI food quality and safety detection. We first introduce the fundamentals of HSI technology and its data processing workflow. We then present a comprehensive and non-mathematical introduction to typical ML and DL algorithms, highlighting their respective strengths and trade-offs. Then, we summarize typical applications across various areas, such as fruit and vegetable quality assessment, meat quality evaluation, egg and tea quality detection, moisture content quantification, variety and origin identification, adulteration and additive detection, heavy metal contamination assessment, mold detection, and plant growth monitoring. Finally, we discuss current challenges, such as data bottlenecks, limited model interpretability and generalizability, and barriers to practical applications, and outline future directions toward intelligent, systematic, and practical HSI systems for food quality and safety detection. Full article

This paper presents a systematic, data-driven literature review of research on Physical Artificial Intelligence (AI) based on the top 100 Google Scholar publications related to the search terms “Physical Artificial Intelligence” and “Physical AI”. The rapid advancement of Physical AI, driven by the convergence of advanced sensor technologies and foundation world models, has resulted in a diverse and fragmented research landscape that lacks comprehensive quantitative overviews. To address this gap, we implement and apply an AI-assisted computational analysis pipeline to this domain. The collected publications are processed using a Large Language Model accessed via a Python-based Application Programming Interface (API), enabling a structured computational analysis of the literature to assist thematic categorization. Based on this approach, the publications are grouped into five data-driven thematic clusters reflecting primary research perspectives within the analyzed sample. Specifically, the identified clusters comprise “Sensor Infrastructure and Architectures”, “Core Learning and Modeling Methodologies”, “Sim-to-Real and Digital Twins”, “Applications”, and “Safety, Governance, and Ethics”. By synthesizing the literature in a structured manner, this work provides a consolidated overview of central research patterns, identifies key operational challenges, and highlights fragmentation across Physical AI research, establishing a solid foundation for future trustworthy autonomous systems. Full article

Accurate prediction of droplet impact behavior is essential for applications including spray cooling, coating technologies, additive manufacturing, and inkjet printing. Conventional analytical and empirical models often have limited predictive capability because of the nonlinear interactions among liquid properties, impact conditions, and surface wettability. This study develops machine learning models to predict the maximum spreading coefficient and the critical spreading–splashing threshold velocity using an experimental dataset of more than 700 droplet impacts spanning multiple liquids and hydrophilic, hydrophobic, and superhydrophobic surfaces. Gaussian process regression (GPR) achieved the highest predictive accuracy, predicting the maximum spreading coefficient with coefficients of determination exceeding 0.99 and outperforming widely used empirical correlations. Evaluation using an externally sourced dataset demonstrated satisfactory model transferability, while a second GPR model accurately predicted the critical spreading–splashing threshold velocity within the investigated parameter space. Shapley Additive Explanations (SHAP) and Individual Conditional Expectation (ICE) analyses showed that impact velocity is the dominant predictor of maximum spreading, whereas surface tension primarily governs splash onset, consistent with established droplet-impact physics. These results demonstrate that interpretable machine learning models provide accurate, physically meaningful predictions across diverse liquid–surface systems, in regimes where the input parameters are not scarcely populated, and offer an alternative to conventional empirical correlations. Full article

Lupin is an increasingly important protein crop, but evidence linking its nutritional and technological potential with fungal contamination, processing safety, and analytical control remains fragmented. This review integrates these areas within a field-to-product assessment of lupin-based food and feed systems, explicitly distinguishing among evidence obtained directly from lupin, findings generated in other legume matrices, and mechanistic inferences. Diaporthe toxica can colonize lupin tissues and seeds, providing a pathway for phomopsins to enter food and feed chains; however, occurrence data are scarce, geographically limited, and focused mainly on phomopsin A (PHO-A). Toxicological evidence is derived predominantly from animal studies, and no health-based guidance values have been established for phomopsins. Processing can reduce quinolizidine alkaloids and other antinutritional constituents, but direct evidence that soaking, heating, fermentation, or germination eliminates phomopsins from lupin is lacking. PHO-A persistence has been demonstrated in artificially contaminated pea-based products. These findings indicate that conventional heating cannot be assumed to detoxify contaminated lupin, but they do not provide quantitative retention data for lupin matrices. LC-MS/MS is currently the most suitable analytical approach, although heterogeneous contamination, insufficiently described sampling, matrix effects, variable sample preparation, and the limited availability of analytical standards and reference materials constrain data comparability. Research priorities include representative occurrence monitoring, validated methods covering relevant lupin matrices, mass-balance processing studies using contaminated lupin, characterization of transformation products, and exposure assessment. This integrated framework distinguishes established evidence from model-derived findings and unresolved knowledge gaps, thereby supporting risk-based control from crop production to final products. Full article

The tourism industry faces increasing demand for personalized travel services alongside environmental sustainability requirements. Recent developments in generative artificial intelligence and large language models provide mechanisms for assisting travelers with itinerary planning, although their integration with tourism data services and sustainability criteria remains under investigation. This paper presents the design and technical evaluation of a conversational agent for sustainability-aware tourism planning in a Technology Readiness Level (TRL) 4 experimental environment. The system combines large language model-based interaction with an external flight information service to generate structured itineraries covering transportation, accommodation, and activities. Sustainability considerations include externally supplied flight emissions information and qualitative recommendation rules for other itinerary components. The controlled evaluation examines functional correctness, natural language processing, external service coordination, response time, and the inclusion of sustainability information. The results indicate that the components can be integrated under the evaluated conditions, while also identifying limitations related to heterogeneous data sources, environmental impact estimation, and the absence of real-world user deployment. The findings concern technical feasibility and do not demonstrate behavioral change or reductions in trip-related emissions. Full article

Aging and cancer form a biological continuum influenced by epigenomic changes, metabolic dysfunction, inflammation, cellular senescence, and loss of tissue homeostasis. Age-related epigenetic alterations can promote cancer, which exploits plasticity for evolution, immune evasion, metastasis, and resistance. Nutrition and metabolism affect this process through one-carbon metabolism, methyl-donor availability, acetyl-CoA and NAD⁺ balance, redox status, microbiome metabolites, and chromatin enzyme activity. Circulating biomarkers such as cell-free DNA methylation, mutation-based ctDNA, fragmentomic features, and non-coding RNAs can detect tumor and host changes linked to aging, inflammation, nutrition, and treatment with minimal invasiveness. This review explores the epigenetic aging–cancer link, how nutrition and metabolism modify pathways, and the potential of circulating biomarkers for diagnosis, prognosis, prediction, and monitoring. The focus is on epigenetic plasticity, drug-tolerant states, resistance, epigenetic drugs, metabolic targeting, and nutritional interventions. New technologies, including single-cell and spatial epigenomics, long-read sequencing, and multimodal computational approaches, aid biomarker discovery and clinical use. Challenges include variability, misclassification, heterogeneity, confounding, reverse causality, overfitting, and limited validation. Clinical applications need standard workflows, representative cohorts, transparent models, and proof that biomarker-guided strategies improve outcomes. Full article

Despite the established clinical benefits of insoluble dietary fibers (IDF), a significant “fiber gap” persists because their inclusion often conflicts with consumer sensory expectations. While general research on dietary fibers is abundant, few studies have systematically isolated the specific impacts of individual components on food quality. To maintain scientific accuracy, we explicitly acknowledge that because the primary literature predominantly evaluates raw agro-industrial by-products (e.g., brans, pomaces, hulls) rather than isolated, chemically pure polymers, a direct, isolated causal relationship to cellulose, hemicellulose, or lignin alone is often confounded by other matrix components. Our framework thus represents a synthesis of the dominant, most plausible roles of these polymers based on converging indirect evidence, rather than causal claims tested on pure substrates. This review characterizes these components through the “Backbone–Matrix–Cement” model to understand their distinct roles in food sensory science. Following PRISMA 2020 guidelines, a systematic search was conducted across ScienceDirect, Web of Science, Google Scholar, and PubMed. The review synthesized data from 106 sources, with 80.2% (85/106) published in the last five years (2020–2026). Data extraction included fiber type, food matrix, and analytical methods, with results grouped thematically. Cellulose (the Backbone) is crystalline, providing mechanical strength and structural stability; it is naturally white and flavor-neutral, primarily influencing firmness and hardness. Hemicellulose (the Matrix) has a high capacity for hydration and acts as a gelling medium. It improves moisture retention and softness in products like bakery goods. Lignin (the Cement) is a rigid, hydrophobic aromatic polymer that is most detrimental to palatability. It consistently causes darkening, bitterness, and astringency. The “coarse granular sensation” or gritty mouthfeel emerged as the primary barrier to consumer acceptance, largely driven by lignin and large cellulose particles. To mitigate these drawbacks, the review identifies several technological interventions: mechanical micronization, biological modifications (such as sourdough fermentation and enzymatic treatments), and chemical modifications (including ozonation and carboxymethylation). For optimal acceptance, formulation levels should generally remain below 10% in bakery products, keeping average particle sizes below the 150–200 µm threshold. Successful development of high-fiber functional foods requires targeted processing strategies that address the specific sensory liabilities of each component while maintaining nutritional efficacy. Full article

To address the challenge of acquiring basketball shooting parameters in a non-contact and quantitative manner, this paper proposes a machine-vision-based prototype system for intelligent basketball shooting feedback. The system constructs a basketball-and-hoop detection dataset for YOLOv5-based object detection, enabling the recognition of basketball and hoop targets in shooting scenarios. By integrating a Semi-Global Block Matching (SGBM) stereo-ranging algorithm, the system obtains the three-dimensional (3D) coordinates of the basketball and the hoop. A shooting evaluation and feedback module is then developed based on an ideal projectile-motion model to analyze the release position, release angle, and release velocity, and to provide reference release velocity, reference force, and make/miss feedback under the ideal model. To improve portability, the complete workflow, including object detection, stereo ranging, and feedback generation, is deployed on an NVIDIA Jetson Nano B01 development board, with TensorRT employed for inference acceleration. Experimental results show that the trained detection model achieves a frame-level recognition rate of about 95% for basketball and hoop detection in the test video after dataset expansion, and that the stereo-ranging module provides effective depth estimation within the scale of the shooting experiments. The system enables trajectory extraction and release-parameter estimation for typical shooting clips. The average relative difference between the velocities extracted by the system and those calculated using the ideal projectile-motion model is 1.96%. In additional experiments involving multiple players, shooting distances, and illumination conditions, the system also shows a certain degree of robustness. Jetson Nano tests further show that the complete TensorRT YOLOv5s and SGBM workflow achieves an average processing rate of 3.65 FPS, an average per-frame latency of 261.31 ms, and an active-stage board power consumption of 5.41 W. Overall, the proposed prototype can provide non-contact, reference-oriented quantitative feedback for basketball shooting training, offering a feasible solution for applying machine vision technology to basketball training assistance. Full article

Accurate acquisition of biomass and yield information for protected vegetable crops is essential for crop growth assessment, environmental regulation, optimal harvest timing, and production planning. With advances in machine vision, spectral sensing, and artificial intelligence technologies, research in this field is shifting from destructive sampling and single-time-point estimation toward non-contact, multisource, continuous monitoring and dynamic prediction. Focusing on protected leafy and fruit vegetables, this review summarizes biomass and yield indicators and their ground-truth measurement methods and compares the characteristics of RGB imaging, three-dimensional vision, spectral sensing, and environmental data. It then reviews advances in biomass estimation, continuous growth monitoring, and harvest prediction for leafy vegetables, as well as flower and fruit sensing, fruit counting, individual fruit mass estimation, and stage-specific harvest yield prediction for fruit vegetables. Current research is expanding from static estimation at the individual-plant level to growth-process monitoring and stage-specific yield prediction, but challenges remain, including interference from complex environments, difficulties in the spatiotemporal alignment of multisource data, incomplete continuous information, and limited model adaptability. Future research should strengthen robust sensing under complex conditions, dynamic multisource fusion, and mechanistic–data-driven integration, thereby advancing the field from individual-plant estimation toward dynamic monitoring of cultivation units and production decision support. Full article

Magnetic cell assembly has emerged as a powerful biofabrication strategy that uses externally applied magnetic fields to manipulate and organize living cells with spatial control, enabling the fabrication of scaffold-free multicellular constructs while preserving cell viability and function. Advances in magnetic nanoparticles, cell labeling techniques, and magnetic field engineering have expanded its applications from rapid spheroid formation to the assembly of complex, spatially organized tissues. This review provides a comprehensive overview of the fundamental principles governing magnetic cell assembly, including the generation of magnetically responsive cells, magnetic force-mediated manipulation, and the biological processes driving tissue formation after magnetic assembly. We discuss the major assembly strategies, including magnetic aggregation, levitation, patterning, alignment, and modular tissue assembly, highlighting their underlying mechanisms, representative studies, engineering advantages, and current limitations. Recent progress in musculoskeletal, cardiovascular, neural, and vascular tissue engineering, as well as organoid and assembloid technologies, disease modeling, and drug discovery, is critically evaluated with an emphasis on experimental outcomes and remaining challenges. Particular attention is given to how magnetic cell assembly has evolved from a technique for manipulating individual cells into a programmable platform for organizing living building blocks with increasing structural and biological complexity. Finally, we discuss the key obstacles to clinical translation, including vascularization, tissue maturation, scalability, reproducibility, and standardization, together with future opportunities arising from the integration of magnetic cell assembly with bioprinting, stem cell engineering, microphysiological systems, and artificial intelligence. This review highlights the potential of magnetic cell assembly as an enabling technology that bridges magnetism and biofabrication to engineer next-generation living tissue models. Full article

Despite technological developments and increasing awareness in the field of occupational health and safety, occupational accidents remain one of the fundamental problems that continue to have significant economic and social consequences. Particularly in the manufacturing sector, the diversity of production processes, differences in the machinery and equipment used, and the complexity of working environments cause occupational accidents to exhibit risk patterns that are both shared across sectors and specific to individual sectors. In this context, this study conducted a topic modeling analysis using the BERTopic algorithm on 1,600,000 occupational accident narratives to identify sector-specific risk patterns and error types associated with 24 different economic activities within the manufacturing sector. To improve topic modeling performance, a multilingual SentenceBERT-based embedding model and a Turkish-specific tr-MTEB embedding model were compared, and the analysis was carried out using the model that provided higher representational capacity and better topic separation performance for the occupational accident narratives. The findings show that, although the sectors differ in terms of production structure, types of machinery and equipment used, material characteristics, and working environments, occupational accidents largely cluster around similar risk mechanisms. As a result, it was determined that the BERTopic method makes the sectoral risk patterns embedded in occupational accident narratives more visible and provides an analytical basis for developing sector-specific prevention strategies in occupational safety practices. Full article

Traditional weight measurement methods need cutting or weighing of the fruit, and this is not practical for preharvest evaluation and market transactions. The suggested approach gives a practical solution for farmers and sellers by giving accurate weight predictions using only external characteristics. Modern technologies, especially artificial intelligence, data analytics and machine learning, are making big changes in the agricultural area. One of the machine learning models used in agriculture is Artificial Neural Networks (ANN). ANN became an important tool in analyzing agricultural data, predicting plant growth processes, finding diseases, determining the effects of environmental factors, reaching productivity goals and detecting weeds and harmful plants. A dataset is created by comprehensively examining 52 watermelons. The dataset includes acoustic properties, geometric measurements, and visual characteristics. The dataset is divided into 40 training samples and 12 test samples. Balanced representation is ensured by using stratified sampling when selecting test samples. The Min-Max normalization method is used for data preprocessing. Nine different training algorithms are comprehensively evaluated within the scope of the study. Eight critical parameters of the ANN models are optimized using four different optimization algorithms to increase the accuracy rate and avoid overfitting. Artificial Bee Colony (ABC), Artificial Fish Swarm Algorithm (AFSA), Whale Optimization Algorithm (WOA) and Grey Wolf Optimizer (GWO) are used as optimization methods. Assessed by five-fold cross-validation, the best configuration (One Step Secant with WOA) achieved a mean absolute error of 0.92 ± 0.28 kg and an RMSE of 1.23 ± 0.37 kg, corresponding to about 10% of the mean fruit weight. Developing a real-time mobile application using the optimized best model will provide practicality in large-scale agricultural enterprises, controlled environments such as greenhouses, and agricultural markets. Full article