Search Results (260)

Plant-derived extracellular vesicles (PDEVs) have emerged as promising natural nanocarriers for biomedical applications owing to their distinctive ability to facilitate intercellular communication and transport bioactive molecules. In this review, we employ bibliometric analysis to identify research hotspots and trends, providing a comprehensive overview of these core themes. The bibliometric results reveal a sustained increase in annual publications in this field, with keyword analysis identifying drug delivery, cross-kingdom regulation, immunomodulation, engineering modification, and gut microbiota as five major research themes. The focus of research has evolved from early basic biological characteristics into engineered smart delivery platforms, with the application areas expanding from intestinal inflammation to neurological, metabolic, dermatological, and oncological diseases. This review systematically examines the core directions in this field. It compares the strengths and limitations of mainstream isolation methods and highlights the value of multi-omics integration, covering the molecular mechanisms of ferroptosis and gut microbiota regulation by PDEVs along with engineering strategies such as drug loading, surface modification, and membrane fusion. It also discusses the latest progress in frontier therapeutic applications of PDEVs, including cancer, inflammatory diseases, tissue regeneration and aesthetics, and neurological disorders. Finally, this review summarizes the key challenges confronting the field, including the lack of standardized protocols, production bottlenecks, and engineering obstacles. It also delineates future directions, including establishing international standardization definitions, advancing multi-omics and AI-driven mechanistic elucidation, developing scalable and efficient purification technologies, and executing systematic preclinical safety and pharmacokinetic evaluations to facilitate clinical translation. Full article

This systematic review summarises advances in artificial intelligence (AI) and machine learning (ML) for foodborne pathogen detection, covering applications in various technologies (AI-assisted microscopy, spectroscopy, biosensors and sensor-based systems), food supply chains, analytical performance, operational metrics and regulatory developments, addressing gaps in previous reviews limited to individual technologies or lacking regulatory analysis. Following PRISMA 2020 guidelines, Scopus, PubMed, and Web of Science were searched from 1 January 2010 to 25 June 2026 using a validated string. Inclusion criteria were explicit detection of a pathogen, clearly described AI/ML algorithm, study evaluation on food or supply chains, and quantitative validation metrics. Exclusion criteria were chemical-only studies, human-diagnostic studies, or purely theoretical studies. Given heterogeneity in the evidence, qualitative quality indicators were favoured over formal quantitative risk-of-bias tools, in distinction to internal cross-validation versus independent external validation. Key data were extracted using a standardised matrix, and after screening and snowballing, the final corpus consisted of 152 studies. CNN (Convolutional Neural Network)-based microscopy provides >99% accuracy in bacterial identification, SERS (Surface-Enhanced Raman Spectroscopy) and CNN 98.68% for pathogens and 99.85% for resistant strains. ML-driven biosensors show 80–100% prediction accuracy in the presence of environmental noise. Yet, performance drops dramatically on external validation, with models falling from 95% internal to 78–82% on independent test sets. Supply chain applications cover meat, dairy, seafood and produce, but most are still at pilot scale. The main constraints are data heterogeneity, lack of public benchmarks, matrix interference, non-standard validation protocols, and regulatory dissonance. However, the integration of AI with Internet of Things (IoT), blockchain and edge computing improves sensitivity, reduces false results and enables real-time monitoring despite the challenges. AI is a powerful decision-support tool that complements existing food safety controls rather than replacing them. To translate these technologies reliably into routine practice, effective implementation requires rigorous external validation and regulatory harmonisation. Full article

Fake news detection has become a major research topic at the intersection of artificial intelligence, data mining, and information security. In this paper, we evaluate the performance of English-trained algorithms on English translations of Romanian-sourced news articles, using a translation-mediated cross-domain evaluation design. The study is based on source code generated with the assistance of artificial intelligence systems for a set of machine learning and transformer-based models. The code was subsequently implemented in Google Colab. 2026, trained on international benchmark datasets, and tested on Romanian news content. This design allowed the rapid prototyping of multiple detection pipelines and the systematic observation of their behavior in a media environment different from that represented in the training corpora. The models were evaluated comparatively using standard classification metrics, including accuracy, precision, recall, and F1-score, complemented by additional indicators relevant to model robustness and practical usability. The experimental results revealed significant differences in performance across algorithms when applied to English translations of Romanian-language news content after training on international datasets. However, this study does not provide a direct comparison between model performance on the international benchmark datasets and the Romanian test corpus; therefore, the gap between the international training corpus and the Romanian-sourced test corpus is interpreted as an exploratory limitation and as a direction for future research. Based on these findings, we propose an empirical classification of the tested models according to their predictive effectiveness, their contextual robustness across linguistic environments, and their operational relevance as filtering tools for institutional monitoring. The results show that AI-assisted coding workflows can provide a viable starting point for reproducible misinformation research, but they also underline the limitations of directly transferring models trained on non-Romanian data to local media ecosystems. The study offers both a replicable evaluation framework and practical insights for institutions involved in strategic communication, public security, and the monitoring of information threats. Full article

Background: Artificial intelligence (AI) is increasingly applied in medical image analysis, although its application in forensic pathology remains limited. The assessment of ligature marks in hanging deaths is challenging and relies on forensic expertise. This pilot study evaluated a deep learning approach for morphological classification of hanging-related ligature marks. Methods: A Convolutional Neural Network (CNN) was trained on a dataset of 404 standardized JPEG images obtained from forensic medicine atlases and classified into hanging-related ligature marks and non-hanging lesions, including strangulation and post-mortem artefacts. Following internal validation, the model was tested on an independent set of forensic case images provided by forensic pathology experts from Messina, Italy, and Vilnius, Lithuania. Images were annotated and reviewed by a team of two forensic pathologists, with final labels assigned by consensus using morphological criteria. Results: The CNN demonstrated encouraging classification performance with an F1-score of 0.81 ± 0.04, distinguishing hanging-related ligature marks from morphologically similar lesions. The methodological framework and image standardization criteria for AI-assisted forensic analysis were also established. Conclusions: AI-based image analysis may support the evaluation of ligature marks during external examinations. Nevertheless, forensic diagnosis requires the integration of autopsy findings, physical examination, and circumstantial evidence. Larger datasets and multicenter protocols are needed to further assess reliability and applicability. Full article

Background: Conversational artificial intelligence (AI), including text-based chatbots, voice-based agents, multimodal systems, and socially assistive robots (SARs), offers a scalable adjunct to therapist-led dementia care. The post-2022 emergence of large language models (LLMs) has accelerated development, yet few reviews apply a unified conversational AI taxonomy across dementia care. This review synthesized the effectiveness, limitations, and implementation challenges of conversational AI across the dementia care continuum. Methods: Six databases (PubMed, Embase, Web of Science, Scopus, IEEE Xplore, ACM Digital Library) were searched for English-language studies (January 2010–March 2026) evaluating conversational AI targeting cognitive, social, or caregiver outcomes. Two reviewers independently screened and extracted data following PRISMA 2020 guidelines; risk of bias used standard tools and findings were synthesized narratively. Protocol: PROSPERO CRD420261333625. Results: Forty studies (8 randomized controlled trials [RCTs], 32 non-randomized) were included. SARs were the largest category (n = 24; 60.0%), followed by text-based chatbots (n = 12; 30.0%), multimodal systems (n = 3; 7.5%), and voice-based chatbots (n = 1; 2.5%). The strongest cognitive evidence came from a social robot RCT (gain of 3.9 points on a 30-point screening measure (p < 0.001). For caregivers, an international RCT (n = 274) showed significant reductions in depression (d = 0.37) and burden (d = 0.34). LLM-based systems produced an 18-fold increase in conversation duration. Speech recognition failure was the most consistently reported technical barrier. Conclusions: Conversational AI shows directional benefit across cognitive, social, and caregiver outcomes. Critical research gaps remain regarding voice-only randomized evidence and adequately powered LLM trials against usual care. Full article

Background: Metallic orthopedic implants face persistent clinical challenges that have proved resistant to incremental conventional development. Machine learning and artificial intelligence offer a complementary paradigm for navigating the high-dimensional design spaces governing implant performance, yet the literature remains fragmented across disciplinary silos with no comprehensive synthesis spanning the full development pipeline. Methods: A structured database search of PubMed/MEDLINE, Embase, and Cochrane (executed May 2026), supplemented by hand-searching of reference lists, identified 33 primary studies organized across five sequential domains: alloy composition discovery, additive manufacturing process–property optimization, lattice and porous structure design, surface engineering and coatings, and corrosion and wear prediction. Results: Across all five domains, machine learning approaches, including random forests, convolutional neural networks, Bayesian optimization, generative adversarial networks, physics-informed neural networks, and autonomous multi-agent platforms, have accelerated property prediction and design space exploration beyond experimental or simulation-based methods. Shared barriers to translation include small, heterogeneous datasets, reliance on internal rather than external validation, limited interpretability, and the absence of regulatory frameworks for AI-assisted device design. Representative performance included modulus predictions within ~4 GPa of first-principles values, ML-designed alloys reaching ~42.7 GPa (versus 103–120 GPa for Ti-6Al-4V), property prediction R² often above 0.90 (up to 0.96–0.9991), 98.3% corrosion severity classification accuracy, and acceleration from a roughly fivefold reduction in finite element simulations to surrogates compressing days into minutes. Conclusions: Addressing these limitations will require open standardized databases linking materials parameters to registry-level clinical outcomes, prospective clinical validation studies, and coordinated engagement between researchers, industry, and regulatory agencies. Full article

Background: Clinical research has evolved into a multidisciplinary field integrating Medical Devices (MD), In Vitro Diagnostics (IVD), and Artificial Intelligence (AI), governed by a modernized European regulatory framework including the Clinical Trial Regulation (CTR), the Medical Device Regulation (MDR), and the In Vitro Diagnostics Regulation (IVDR). In 2021, the Istituto Oncologico Veneto (IOV) IRCCS established the Clinical Trial Quality Team (CTQT) to provide support for non-profit trials and ensure high-quality standards. Methods: A descriptive analysis of trials managed between 2021 and 2025 was conducted using the REDCap platform. A team of 15 professionals assessed performance via Key Performance Indicators (KPIs) categorized into: ethical–regulatory compliance, operational efficiency and data quality. Results: The CTQT manages 18 studies (83% interventional, 50% multicentre), primarily focused on brain (17%) and genitourinary (23%) tumours. The mean time from internal feasibility to Ethics Committee discussion is 13 days. Total approval time (ethical and administrative) is 107 days. Operational metrics are strong, with Site Initiation Time averaging 27 days and First Patient In (FPI) at 41 days. Conclusions: A centralized, multidisciplinary structure effectively supports non-profit research in a complex regulatory environment. While operational speed is high, challenges such as staff turnover and query management variability remain. Future efforts will focus on standardizing post-approval administrative phases and optimizing data management to further improve trial efficiency and integrity. Full article

Despite the rapid expansion of unmanned aerial vehicle (UAV) applications across precision agriculture, logistics, infrastructure inspection, disaster response, and aerial surveying, objective and quantitative hardware evaluation criteria for UAV components remain insufficiently developed. This paper proposes quantitative key performance indicators (KPIs) for thirteen core hardware subsystems, including airframe and propulsion, battery and power supply, flight control, wireless communication, imaging (camera), Global Positioning System (GPS)/Global Navigation Satellite System (GNSS) positioning, thermal management, acoustic and vibration characteristics, AI-based autonomous flight, electromagnetic compatibility (EMC), cybersecurity, and reliability and environmental qualification, together with LiDAR payload evaluation criteria. International standardization activities by 3GPP (Release 15/17), IEEE (1936–1958 series), American society for photogrammetry and remote sensing (ASPRS), and national regulatory frameworks are synthesized to define measurable performance metrics and recommended test methods for each subsystem. An integrated KPI matrix maps application-domain-specific performance targets—encompassing surveying (real-time kinematic (RTK) horizontal accuracy ≤ 2 cm root-mean-square error (RMSE), ground sample distance (GSD) ≤ 2 cm/px), infrastructure inspection (LiDAR payload up to 8 kg, beyond visual line-of-sight (BVLOS) latency ≤ 140 ms), and logistics delivery (payload ≥ 2 kg, precision landing ≤ 50 cm)—demonstrating that no universal platform can simultaneously satisfy all domain requirements. A fuzzy-AHP weighting procedure and inter-subsystem coupling analysis are introduced to address size, weight, and power (SWaP) trade-off relationships that purely additive scoring models cannot capture. The proposed evaluation framework is intended to contribute practically to UAV standardization, certification, and quality management across the full design–procurement–operation lifecycle. Full article

Fundus fluorescein angiography (FFA) is the gold standard for diabetic retinopathy (DR) staging, yet whether angiographic phase affects deep learning performance remains unknown. Using 7508 FFA images from 863 eyes, stratified into venous, recirculation, and late phases, we developed Swin Transformer- and ConvNeXt-based DR staging models under the International five-grade and Chinese six-grade classification systems. Multiclass and binary (NPDR vs. PDR) classification tasks were evaluated. This study provides the first systematic quantification of angiographic phase effects on FFA-based DR staging. Phase-related performance was analyzed using generalized linear mixed-effects models with beta regression. ConvNeXt generally outperformed Swin Transformer, particularly in multiclass classification. Across all experimental settings, performance showed a consistent numerical decline from the venous to the late phase. Under the International five-grade system, ConvNeXt accuracy declined from 86.67% to 81.87%; however, Bonferroni-adjusted comparisons revealed no statistically significant phase-related differences (all adjusted p > 0.05), with small effect sizes (|SMD| < 0.2). Binary classification remained highly stable across phases, with accuracies exceeding 91%. Grad-CAM visualizations demonstrated progressively diffuse model attention in later phases. These findings support phase-flexible FFA acquisition for binary DR screening, whereas venous- or recirculation-phase images remain preferable for high-precision multiclass staging, guiding phase-aware AI development. Full article

Urban centers today face unprecedented challenges in energy management, emissions, waste disposal, and public services, as nearly 70% of the global population is projected to live in cities by 2050. The complexity and rapid evolution of urban systems underscore the pressing need for stability, innovation, and adaptability, particularly regarding sustainability and digital transformation. AI-powered simulation models have emerged as transformative tools, capable of simplifying, predicting, and managing highly intricate urban systems while offering policymakers valuable insights for strategic planning. However, the integration of AI in smart and sustainable urban development presents critical legal, ethical, and regulatory concerns. This study examines these questions by evaluating existing and emerging frameworks addressing algorithmic transparency, data protection, stakeholder engagement, and sustainable development in prominent urban models including Amsterdam, Copenhagen, Singapore, Tokyo, Bangalore, and Nairobi. Comparative analysis is conducted through a doctrinal desk review, focusing on statutory provisions, international policies (EU, UN-Habitat), ISO Smart City standards, and local governance charters. Key issues addressed include the risk that AI models, if unregulated, become opaque “black boxes” that obscure both decision-making logic and accountability. Without robust standards, there is no guarantee of interoperability, revision, or representation of public interest. Equitable management, access, and inclusive participation are vital to responsible AI frameworks in urban planning. This article advances a comprehensive legal and policy framework for ensuring accountability, transparency, and stability in AI-driven city governance, bridging gaps between technological innovation, urban studies, and regulatory oversight. The proposed governance structure empowers cities to adopt multi-level, authority-driven mechanisms that safeguard the common good while leveraging AI’s potential in sustainable urban transformation. Full article

Background/Objectives: Peri-implantitis is a major complication in implant dentistry, and its radiographic diagnosis remains challenging because conventional assessment is operator-dependent and bone loss is often detected only after measurable changes occur. Artificial intelligence (AI) may support the detection, quantification, and prognostic assessment of peri-implant bone conditions. This review aimed to synthesize evidence on AI-based radiographic approaches for peri-implantitis detection, marginal bone loss measurement, and risk stratification. Methods: PubMed and Scopus were searched for original studies published between 2013 and 2025 that applied artificial intelligence (AI), including machine learning and deep learning, to peri-implantitis. Eligible studies focused on peri-implant bone assessment and reported quantitative performance metrics. Extracted data included imaging modality, AI model, task, dataset, reference standard, validation strategy, performance, and clinical relevance. A qualitative synthesis was performed. Results: Eleven studies met the eligibility criteria; however, one full text could not be retrieved, and ten studies were included. In most of the studies, peri-implant marginal bone loss detection or measurement was performed using periapical/intraoral radiographs, while only few studies used panoramic or combined imaging. Common architectures included YOLO variants, Faster R-CNN, Mask R-CNN, U-Net, ResNet, and AlexNet. Performance was generally encouraging for implant localization, bone loss detection, keypoint identification, and severity classification. Only one study addressed outcome prediction. All studies were retrospective and internally validated. Conclusions: AI may support radiographic detection and quantification of peri-implant bone loss as an adjunctive diagnostic tool. However, evidence is limited by retrospective designs, heterogeneous reference standards, lack of external validation, and limited clinical-data integration. Future studies should prioritize prospective multicenter validation, longitudinal imaging, and multimodal models. Full article

Robo-advisors have expanded access to automated investment services, but many platforms continue to rely on relatively static onboarding procedures and limited forms of user interaction. This study examines how participants with investment experience respond to two next-generation robo-advisory design features: financial digital twins, understood as dynamic investor profiles that integrate goals, risk tolerance, cash-flow patterns, and anticipated life events, and conversational artificial intelligence (AI), understood as an interactive interface for explaining recommendations. Using a scenario-based randomized 2 × 2 online experiment, 336 adult respondents with self-reported investment experience, recruited through professional and academic networks, were assigned to one of four robo-advisor scenarios that varied the personalization architecture, standard profile versus digital twin, and the interface style, plain dashboard versus conversational AI, while holding the portfolio recommendation constant. The results show that digital-twin personalization increases perceived personalization and privacy concern, indicating that more adaptive advisory architectures may be viewed as both more relevant and more data-intensive. Conversational AI increases the perceived interactive quality of the advisory experience, while selected willingness-related patterns, especially in the combined digital-twin and conversational-AI condition, are treated as exploratory because several secondary composites displayed limited internal consistency. The strongest confirmatory emphasis is therefore placed on perceived personalization and privacy concern, and the remaining findings are best interpreted as scenario-based investor responses rather than evidence of actual adoption behavior or confirmed psychological mechanisms. The study contributes to behavioral FinTech research by clarifying the personalization–privacy tension in AI-enabled robo-advisory services and by offering design implications for more transparent, interactive, and responsibly personalized digital wealth-management systems. Full article

Plasma electrolytic oxidation (PEO) has emerged as a leading surface engineering technology for improving the tribological and corrosion performance of lightweight structural alloys, including aluminum, magnesium, titanium, and zirconium. Unlike conventional anodizing or line-of-sight deposition processes, PEO forms thick, multiphase ceramic oxide coatings metallurgically bonded to the substrate through plasma-assisted in situ oxidation, enabling treatment of complex and internal geometries that competing technologies cannot reach. The tribological performance of PEO coatings is governed by coupled interactions among electrolyte chemistry, electrical discharge behavior, phase evolution, porosity development, and residual stress state. This review critically evaluates the friction, wear, and tribo-corrosion behavior of PEO coatings under dry sliding, lubricated, high-temperature, marine, and vacuum environments, and systematically examines the influence of processing parameters, microstructural evolution, transfer layer formation, and counterface interactions on coating performance. Hybrid and duplex systems incorporating solid lubricants, polymer impregnation, sol–gel sealing, and multilayer architectures are discussed as strategies to overcome limitations associated with brittleness and surface porosity. Current research challenges, including fatigue degradation, coating defect control, limited cross-study standardization, and incomplete mechanistic understanding of process–microstructure, tribological relationships, are critically assessed. Emerging directions encompassing self-lubricating adaptive coatings, AI-guided process optimization, and multifunctional hybrid architectures are highlighted as pathways toward next-generation surface systems. This review provides a mechanism-based framework for understanding tribological behavior in PEO coatings and identifies critical opportunities for future industrial implementation in aerospace, automotive, marine, biomedical, and energy applications. Full article

Background/Objectives: Accurate staging of mediastinal lymph nodes is essential for therapeutic decisions and prognostic assessment in non-small cell lung cancer (NSCLC). This systematic review evaluates diagnostic performance, validation strategies, and clinical significance of artificial intelligence (AI)-based 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET/CT) radiomics models for mediastinal nodal staging in NSCLC. Methods: Systematic literature searching was conducted in PubMed, ScienceDirect, and Scopus according to the PRISMA 2020 guidelines. Eligible studies used radiomic or AI-based approaches for mediastinal lymph node (LN) evaluation in NSCLC, with histopathology as a reference standard. Extracted data included study design, cohort characteristics, imaging method, validation strategy, and diagnostic performance metrics. Methodological quality was assessed by the QUADAS-2 tool. Results: Thirteen studies were included which are mainly retrospective in their designs with cohort sizes varying between 87 and 3265 patients. Models evaluated on external or prospective validation cohorts generally showed lower performance compared with training or internal datasets. However, clinically significant discriminative ability has been preserved across heterogeneous populations. In studies that directly compared methods, composite models integrating radiomic features with clinical factors and conventional PET metrics, sometimes including deep learning-derived features, consistently outperformed radiomics-only models. Additionally, selected approaches addressing FDG-related false-positive uptake improved distinction between benign and metastatic mediastinal lymph nodes; this is reflected by reduced false-positive classifications plus higher specificity compared with conventional PET/CT interpretation. Conclusions: AI-based 18F-FDG PET/CT radiomics show a promising discriminative capacity for mediastinal nodal staging in NSCLC, especially when it is integrated with clinical and conventional imaging variables. Although the model performance remains clinically significant within independent validation cohorts, attenuation compared with training datasets is commonly observed. Methodological heterogeneity, predominantly retrospective study designs, and the scarcity of prospective multicenter validation currently limit routine clinical implementation. Full article

The rise of artificial intelligence (AI) and classroom automation demands rethinking visual and arts education. To prevent learning standardization, it is imperative to cultivate a critical teacher profile capable of leading new digital ecologies. In this context, measuring the creative self-perception of future educators constitutes a fundamental pedagogical need. This instrumental study analyses the factor structure and internal consistency of the Creative Personality Scale (CPS), adapting it to contemporary technological challenges. It was administered to 90 pre-service teachers from the Early Childhood and Primary Education programmes at the University of Almería. Through an Exploratory Factor Analysis (EFA) using Principal Axis Factoring (PAF) with Oblimin rotation, the scale was refined to 17 items, confirming a robust three-dimensional structure: Imaginative Creativity, Behavioural Originality, and a Positive Attitude towards Challenges (explaining 44.17% of the variance; α = 0.862). While not a direct measure of pedagogical performance, these dimensions capture the psychological dispositions hypothesized as necessary for educators to critically navigate AI integration and mitigate algorithmic standardization. In conclusion, the adapted scale provides an initial exploratory validation of a diagnostic framework. Its application provides a foundational metric for teacher education programmes, aiming to foster learning environments where technology integration is deliberately guided by human judgment and sensitivity. Full article