Search Results (1,754)

The integration of Artificial Intelligence (AI) into Internet of Things (IoT) ecosystems has enabled the development of advanced cyber–physical systems, including smart appliances, while introducing security, privacy, and AI governance risks that extend beyond the scope of traditional threat models. Existing approaches often address cybersecurity, AI risk management, and regulatory compliance in isolation, leaving manufacturers without a systematic method for translating identified threats into architectural controls and certification requirements. To address this gap, this study proposes a Security-by-Design and risk-based certification framework that combines a six-layer IoT-AI reference architecture with STRIDE-based threat analysis augmented to capture AI-specific threats, including prompt injection and data poisoning. The resulting cross-layer analysis informs a four-level certification model (L1–L4) that deterministically maps each appliance configuration to a set of mandatory security and governance controls according to its degree of autonomy and AI capability. The framework is instantiated and evaluated using a physical smart-refrigerator prototype, demonstrating how threat identification can be systematically translated into design decisions and certification requirements. The proposed framework provides manufacturers, certification bodies, and researchers with a reproducible engineering pathway for designing and evaluating secure, governance-aligned AI-enabled IoT appliances. Full article

Robotic joint motor drive systems must combine torque density and dynamic response with low mechanical impedance, safe interaction, and thermal robustness. This review treats gear ratio as a system-level design coordinate realized jointly by the motor, transmission, thermal path, sensing, and control. It synthesizes how ratio selection changes torque–speed capability, reflected inertia, losses, thermal duty, reducer nonidealities, backdrivability, and control bandwidth. The proposed spectrum uses nominal ratio as its primary coordinate while treating reducer topology, application domain, integration level, and compliance as distinct, overlapping descriptors. Mechanism-level conclusions are based on peer-reviewed studies; manufacturer specifications, open-source structures, and model-based engineering examples are identified and interpreted within narrower evidence boundaries. Representative robotic-joint cases connect these mechanisms to application demands, and an iterative framework translates the synthesis into checks on the task envelope, motor–reducer matching, thermal feasibility, transmission nonlinearity, sensing, and control. Relative to gearbox-centered reviews and task-specific motor–transmission optimization studies, this review provides a cross-domain decision map rather than a product ranking or universal predictive model. Full article

Generative artificial intelligence (GenAI) is being absorbed into education as a new infrastructural layer, promising individualized tutoring, instant feedback, translation, and accessibility support at marginal cost. Sustainable Development Goal 4 (SDG 4) calls for inclusive and equitable quality education for all, yet the evidence on whether GenAI advances or undermines that goal points firmly in both directions at once. At the task level, GenAI and intelligent tutoring systems repeatedly compress performance distributions, with the largest gains accruing to lower-skilled and lower-baseline participants. At the system level, a parallel literature on access, AI literacy, language, disability, teacher capacity, and over-reliance finds that benefits are conditioned by resources that track prior advantage. These two literatures are usually read as being in tension, and the tension is usually resolved by privileging one of them. This article argues that both are correct and that the appearance of contradiction is an artifact of conflating distinct stages of a single pathway. We develop the divide cascade: a four-stage filter model—access, effective use, benefit realization, and durable learning—in which each stage has a pass rate that correlates with prior advantage. Because pass rates compound multiplicatively across stages while compression acts additively within a stage, a technology can compress outcomes among those who clear every filter and still stratify outcomes across the population as a whole. We formalize this structure, derive the condition under which the stratifying force dominates the equalizing one, and state three predictions that distinguish the cascade from an access-centred account: that access-only interventions should attenuate rather than close benefit gaps, that a single intervention can narrow one gap while widening another simultaneously, and that measured equity gains should decay as the evaluation horizon lengthens. We also specify what would falsify the model. The framework is then used to identify the conditions that set the sign of GenAI’s distributional effect, to map those conditions onto SDG 4 targets, and to derive a testable research and policy agenda. A recurring corollary is methodological: the strongest evidence for compression comes from workplace-productivity studies that measure produced artifacts rather than durable learning, so its transfer to education is an open question that the cascade locates precisely rather than assumes. GenAI, we conclude, is neither inherently an equalizer nor an amplifier; it is a multiplier whose sign is set by how completely the cascade is engineered for the learners who start behind. Full article

Background/Objectives: Immune-mediated inflammatory skin diseases (IMSDs), including psoriasis, atopic dermatitis, and systemic sclerosis, are chronic relapsing disorders that impose substantial physical, psychological, and economic burdens. Current assessment relies largely on subjective clinical scoring and conventional imaging modalities that provide limited functional information. This structured narrative review evaluates photoacoustic imaging (PAI) as a non-invasive optical-ultrasound technique for morphological and functional assessment in IMSDs. Methods: We searched PubMed studies published from January 2017 to December 2025 using terms related to photoacoustic imaging, optoacoustic mesoscopy, psoriasis, atopic dermatitis, systemic sclerosis, contact dermatitis, and immune-mediated inflammatory skin disease. Studies were narratively synthesized by disease, PAI modality, comparator, outcome, and translational limitation. Main Findings: PAI can map vascular and pigment-related optical absorption and estimate oxygenation, blood volume, and selected structural biomarkers. In psoriasis, atopic dermatitis, and systemic sclerosis, PAI-derived measures have been associated with clinical severity, subclinical vascular or epidermal changes, and treatment response. However, most available studies are small, single-center, heterogeneous, or based on prototype systems, and several biomarkers still require histopathological and external validation. Conclusions: Current evidence supports PAI as a feasible research and potential adjunctive assessment tool for IMSDs rather than an established replacement for standard clinical, histopathological, or imaging methods. Standardized protocols, validation across skin phototypes, cost-effective devices, and prospective multicenter studies are required before routine clinical adoption. Full article

In confined waters such as cylindrical aquaculture cages and ponds, underwater robots for cleaning, harvesting and other tasks that use single-beam mechanical scanning sonar for positioning and navigation are susceptible to multipath interference in complex water environments. Meanwhile, under extreme sea conditions, the severe attitude swaying of the robot and the slow-scanning characteristic of the sonar superimpose on each other, causing range stretching and helical deformation of the acoustic point cloud. To address these problems, this paper analyzes the deformation mechanism of single-beam sonar and proposes a spatiotemporal joint deformation compensation and localization-mapping method. First, an attitude-derived probabilistic confidence model is introduced as a lightweight robustness safeguard to characterize the geometric reliability of sonar echoes and reduce the contribution of low-confidence measurements during subsequent registration. Second, a beam-level spatiotemporal joint de-deformation algorithm is designed: the slant range in polar coordinates is flattened to eliminate nonlinear swaying deformation, and a beam-level displacement back-estimation based on the beam time offset and feedback velocity is employed to remove helical misalignment, thereby enhancing the underlying correction capability for dynamic deformation processes. Finally, a lightweight SLAM architecture that integrates keyframe-based dynamic sub-maps is constructed, where a confidence-weighted ICP is used to estimate the planar position with the heading provided by the compass and provide velocity-based closed-loop feedback, effectively mitigating the problem of global matching divergence caused by underlying dynamic deformations. Real-data-driven semi-physical disturbance tests based on measured pool data show that, under the injected ±45° roll disturbance and translational drift, the proposed method reduces the maximum point-to-reference error MaxAE from 1.059 m to 0.098 m. The reported mean internal registration residual decreases from 0.275 m for the traditional navigation odometry SLAM to 0.158 m for the proposed method, corresponding to a numerical reduction of approximately 42.5%. Under the evaluated conditions, the proposed method effectively mitigates point-cloud deformation and registration instability caused by robot swaying and slow-scanning sonar, while confidence weighting is retained as an auxiliary robustness mechanism for handling low-confidence correspondences. Full article

High shrinkage-related defect rates in aluminum die casting reduce effective production capacity. They also create energy-intensive re-melting loops, which weaken production planning reliability and environmental performance. This study examines how Green Lean Six Sigma can stabilize a resource-constrained Small and Medium-Sized Enterprise (SME) die-casting process and translate quality improvement into measurable capacity recovery and Scope 2 electricity-related carbon savings. Based on a 10-month case study, the Define–Measure–Analyze–Improve–Control (DMAIC) framework was integrated with factorial ANOVA, the Response Surface Methodology (RSM), one-way analysis of variance (ANOVA) and statistical process control (SPC). These methods supported process parameter identification, operating-window development and shop-floor process stabilization. The analysis identified the filling speed and mold temperature as significant shrinkage drivers, developed a mold temperature control map, and determined the standardized filling speed at 800 mm/s. The intervention reduced the shrinkage defect rate from 7.19% to 1.46%, reduced the overall scrap rate from 7.60% to 2.71%, and improved the overall sigma level from 2.93 to 3.42. This yield improvement generated a 4.89 percentage-point yield-equivalent capacity gain, avoided 4401 kWh of re-melting electricity, reduced Scope 2 emissions by 2.36 t CO₂e, and generated gross annualized savings of RMB 249,214 (USD 35,783). Considering a one-time implementation cost of RMB 31,445 (USD 4515), the first-year net saving was RMB 217,769 (USD 31,268). The findings show that accessible statistical process control methods can provide SMEs with a resource-efficient pathway to improve process stability, capacity utilization and electricity-related environmental performance before investing in advanced digital technologies. Full article

This article presents a practice-oriented methodological proposal for designing inclusive multimodal learning content with Generative AI (GAI) in migrant adult literacy. It does not report an experimental intervention or a statistical evaluation of learning outcomes. Its contribution lies in formalizing a context-sensitive design pathway for early preA1–A2 literacy and language-learning provision in Italian CPIA settings, where learner profiles are highly heterogeneous, attendance may be discontinuous, and written language is both a learning goal and a barrier to participation. Unlike generic AI-supported instructional design frameworks, the proposed approach starts from recurrent communicative needs in adult migrant education and translates them into short, modular and reusable learning artifacts that coordinate textual, visual, audio-oral and interactive layers. The framework distinguishes multimodal design, understood as the pedagogical coordination of different semiotic modes, from the mere use of multiple media. It also integrates accessibility as a set of concrete design criteria, including linguistic readability, visual clarity, audio quality, layout, font size, contrast, cognitive load and usability in print or mobile formats. The article outlines a sequence of design operations: mapping learner profiles, selecting situated communicative scenarios, generating and revising textual material, developing visual and audio scaffolds, structuring guided interaction, and applying pedagogical, cultural and ethical review. An illustrative micro-unit on asking for information at a municipal office shows how this pathway can support dialog, visual glossary, audio practice, role-play and formative assessment. The proposal is intended for CPIA educators, adult literacy professionals, instructional designers and researchers in multimedia learning and educational technology. Its educational implication is that GAI can support inclusive material design only when its outputs are treated as provisional resources to be selected, adapted and validated through human pedagogical judgment. Full article

Architectural heritage interventions often preserve visual form while leaving acoustic memory and everyday sound practices outside the design brief. This article positions soundscape as an architectural material for heritage media architecture and tests it through Sanggar Kungfu Kapasan, a historic Chinese-diaspora kampung in Surabaya, Indonesia. The research combined contextual and soundmark mapping, Pepper’s Ghost prototyping, architectural translation of a semi-transparent virtual layer, and comparative evaluation using 2D visualization and immersive VR. After deduplication, 37 participants evaluated the 2D material, 31 evaluated VR, and 24 completed both modes for within-subject analysis. VR significantly improved perceived physical quality (Δ = +0.46, p = 0.0007, d_z = 0.80), communication quality (Δ = +0.42, p = 0.0218, d_z = 0.50), and overall evaluation (Δ = +0.20, p = 0.0285, d_z = 0.48). Sound-specific VR ratings were positive for place-meaning support (M = 5.87) and scene formation (M = 5.65), and the soundscape design index correlated with communication quality (r = 0.47, p = 0.007). The findings show that soundscape strengthens narrative legibility and architectural fit when deliberately coordinated with visual layering, scene sequencing, and cultural context. The study contributes a soundscape-informed, feng shui-readable workflow for pre-installation evaluation of multisensory heritage interventions. Because the respondent pool was predominantly composed of architecture students, the results are interpreted as a preliminary design-user evaluation rather than community validation. Full article

Background: Patient-specific targeted radionuclide therapy (TRT) requires consideration not only of administered activity but also of the spatial distribution of radiopharmaceutical uptake and radionuclide-specific energy deposition. This study developed a voxel-based computational workplan to compare ¹⁷⁷Lu, ¹⁶¹Tb and ⁹⁰Y, together with hybrid radionuclide models, using patient-specific PSMA PET-derived tumour activity distributions. Methods: PSMA PET/CT data from 20 patients with prostate cancer, comprising 10 ¹⁸F-PSMA and 10 ⁶⁸Ga-PSMA examinations, were processed to obtain 2285 quality-filtered lesions. Radionuclide-specific dose-point kernels (DPKs) were generated in water using OpenGATE and applied to voxel-wise lesion activity distributions to reconstruct absorbed-dose maps. Kernel characteristics were evaluated using radial energy-containment metrics, and ¹⁷⁷Lu, representing ¹⁶¹Tb simulations, was subjected to grid-convergence testing and external comparison with a published DPK. Lesion dosimetry was assessed using Dmean, D90, D95, equivalent uniform dose (EUD) and tumour control probability (TCP), with uncertainty quantified using patient-cluster bootstrap confidence intervals. Kinetic sensitivity and diagnostic tracer subgroup analyses were additionally performed. Results: The study showed that ¹⁶¹Tb produced the highest median lesion-level Dmean, D90, D95 and EUD at 182.79, 136.28, 132.07 and 148.53 Gy, respectively, with a median TCP of 0.981. Corresponding values for ¹⁷⁷Lu were 141.33, 105.42, 102.10 and 114.78 Gy (TCP 0.930), while ⁹⁰Y produced lower local dose metrics but the broadest radial dose distribution, consistent with its longer-range β-particle crossfire. ¹⁶¹Tb remained the highest-ranking radionuclide across the investigated kinetic cases and within both diagnostic tracer subgroups. Hybrid ¹⁶¹Tb/⁹⁰Y kernels provided a controllable compromise between localised energy deposition and extended crossfire; a 70:30 model increased central dose localisation while retaining an R90 and R95 of 6 and 7 mm, respectively. Grid-convergence and published-DPK comparisons supported the numerical adequacy of the kernel methodology. Radionuclide emission characteristics substantially influence the transformation of heterogeneous tumour uptake into spatial absorbed-dose distributions. Within this model, ¹⁶¹Tb provided the strongest overall lesion-level dosimetric performance, whereas the extended range of ⁹⁰Y may offer complementary crossfire for selected bulky or heterogeneous lesions. Conclusions: The findings support phenotype-informed radionuclide comparison and provide a computational basis for investigating hybrid strategies. However, the absolute dose estimates and proposed radionuclide combinations remain model-based and require validation using serial therapeutic imaging, heterogeneous patient-specific dosimetry and normal-organ dose constraints before clinical translation. Full article

Lead remains a public health priority, in spite of long-standing mitigation efforts. In drinking water, the source of lead is primarily the domestic distribution system. Thus, regulatory compliance monitoring fails to capture the magnitude of the problem. In order to assess lead concentrations in drinking water and identify populations at risk of lead exposure via drinking water, a nationally representative survey was carried out in Hungary. Municipalities were categorized and, where necessary, subdivided into neighborhoods based on the type of municipality, the corrosive nature of the supplied drinking water, and the potential presence of lead pipes in the mains and the domestic distribution systems. A total of 58 representative residential areas were selected for a 6-month sampling campaign. Random daytime (RDT) and 1 min flushed samples were collected from residential and public buildings and analyzed for lead by inductively coupled plasma mass spectrometry. Non-compliance rates were significantly higher in buildings built before 1945 (25% compared to 7% in newer buildings), when the use of lead pipes was general practice. Flushing reduced lead levels significantly, corresponding to 14% and 6% non-compliance in the RDT and flushed samples, respectively. No association was found with corrosive versus protective water quality. High risk areas were primarily identified in the capital and historic centers of cities. The results were translated into public information tools including a searchable map of lead risk areas and a simplified risk assessment tool for consumers. The survey outcomes also support national risk assessment for lead in drinking water, an upcoming obligation under the recast European Union drinking water regulation. Full article

Monocular camera relocalization recovers a six-degree-of-freedom pose from one RGB image, but direct absolute pose regression typically predicts translation and rotation from one terminal representation. We propose Decoupled SwinPose, a hierarchical Swin-Tiny regressor that instead reads translation from a shallow, higher-resolution Stage 1 map and rotation from the deep, contextual Stage 3 representation, testing this asymmetric-readout hypothesis through three falsifiable predictions. Across three TUM RGB-D motion regimes and all seven Microsoft 7-Scenes environments, against constant-pose, retrieval, and matched shared-terminal controls, it achieves the lowest three-seed mean translation and rotation error on all ten evaluated sequences within the matched reimplemented cohort. Relative to the strongest alternative learned model within this cohort, translation error falls by up to 29.6% on TUM RGB-D and 52.6% on 7-Scenes, and rotation error by up to 27.2% and 54.4%, respectively. Routing ablations, including a capacity-matched control, support T1-R3 as the strongest overall trade-off among six tested configurations, and matched profiling shows the readout adds only 0.29% parameters with essentially identical FLOPs and FP32 cost relative to a shared-terminal baseline on an NVIDIA L4. These results support asymmetric multilevel readouts as an effective, low-cost architectural prior for transformer-based monocular pose regression in the evaluated indoor setting. Full article

Public parks support health, wellbeing, and social inclusion, yet conventional wayfinding approaches remain largely focused on physical mobility and directional navigation, often overlooking the cognitive and sensory experiences of neurodivergent users. This study develops a theory-informed framework for neuroinclusive wayfinding design through an integrative literature review. Using a Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA)-informed protocol, 106 core sources were synthesized from landscape architecture, environmental psychology, public health, and neurodiversity research. A qualitative theory-building approach combining thematic analysis, theoretical mapping, and abductive synthesis identified four themes: Cognitive Navigation and Environmental Legibility, Sensory Processing and Environmental Regulation, Restorative Landscapes and Nature-Based Wellbeing, and Inclusive, Adaptive, and Neurodiverse Design. These themes were subsequently translated into eight environmental challenges and operationalized into eight neuroinclusive wayfinding design criteria organized within four support domains. The resulting framework conceptualizes wayfinding as a cognitive–sensory support system rather than a purely navigational function, linking interdisciplinary evidence to practical landscape architectural design strategies. By integrating fragmented knowledge across multiple disciplines, the study provides a foundation for future validation and the advancement of neuroinclusive public park design. Full article

Conventional heavy metal remediation technologies are constrained by low efficiency, secondary pollution risks, and limited scalability. Dissolved organic matter (DOM), with its green, cost-effective complexation properties, has become a promising pathway for pollution control. Existing patent bibliometric studies in this field suffer from single-database bias, limited causal quantification of policy impacts, and incomplete depiction of global–local technological heterogeneity. To address these gaps, this study maps the technological innovation landscape of DOM interactions with four typical heavy metals (Cd, Pb, Cu, Zn) during 2004–2024, using a complementary dual-database framework combining Derwent and IncoPat. We integrate a three-dimensional “time–region–technology” analytical framework with interrupted time series analysis (ITSA), after standardized data processing including family deduplication and citation normalization. Cross-validation confirms that China contributes the largest share of global patent output (46.6% in Derwent, 55.0% in IncoPat). Three milestone environmental policies in China exert sequentially intensifying causal effects on patent growth (all p < 0.05), forming a closed-loop mechanism of policy orientation, funding support, technology transfer, and international diffusion. We identify a pronounced global–local technological divergence: global frontier innovation centers on digital basic research, whereas local innovation in China prioritizes engineering applications. Core patents advance the field through cross-domain technology adaptation, and the representative technical paradigm (exemplified by patent CN101168852A) has been industrially validated. These findings provide empirical support for engineering translation and policy optimization in DOM-based heavy metal remediation. Full article

Background: Memantine, an uncompetitive and voltage-dependent N-methyl-D-aspartate (NMDA) receptor antagonist, is clinically established for moderate-to-severe Alzheimer’s disease. Its pharmacodynamic profile, low-to-moderate affinity, rapid open-channel block, and strong voltage dependency allows selective inhibition of pathological NMDA overactivation while preserving physiological neurotransmission. Increasing evidence shows that these same mechanistic principles operate in the peripheral nervous system, where NMDA receptors contribute to excitotoxicity, oxidative stress, neuroinflammation, and maladaptive nociceptive signaling. Purpose: To synthesize emerging preclinical and clinical evidence demonstrating memantine’s modulatory and neuroprotective actions in peripheral neurons and glia and to outline implications for drug repurposing across neurology, pain medicine, oncology, supportive care, and ophthalmology. Methodology: A narrative integration of mechanistic studies, in vivo preclinical models, and heterogeneous clinical trials evaluating memantine’s effects on peripheral sensory neurons, autonomic neurons, Schwann cells, retinal ganglion cells, and neuromuscular junction physiology. Evidence was examined across conditions involving excitotoxicity, oxidative injury, mitochondrial dysfunction, apoptotic signaling, neuroinflammation, and neuropathic pain amplification. Results: Memantine consistently attenuates peripheral excitotoxic calcium influx, suppresses NOX-2–mediated ROS generation, stabilizes mitochondrial membrane potential, modulates Bax/Bcl-2 signaling, and reduces neuroinflammatory cytokine activity. It also inhibits dorsal horn wind-up selectively under neuropathic conditions. These convergent mechanisms yield protective effects across chemotherapy-induced peripheral neuropathy (CIPN), diabetic neuropathy, traumatic nerve injury, phantom limb pain, retinal ganglion cell excitotoxicity, and organophosphate-induced neuromuscular toxicity. Clinical evidence includes improved multimodal neuropathy outcomes in diabetic neuropathy when combined with gabapentin, reduced phantom limb pain prevalence and intensity at six months, and a five-fold reduction in post-mastectomy neuropathic pain with pre-emptive administration. Conclusions: Memantine should be conceptually reframed as a system-wide neuroprotective agent with substantial translational potential beyond the CNS. Priorities for future development include NR2B-selective peripheral NMDA antagonists, peripherally restricted formulations, single-cell transcriptomic mapping of peripheral NMDA receptor subtypes, and adequately powered PNS-specific randomized trials. Full article

We present ESNformer, a hybrid architecture that couples an Echo State Network (ESN) reservoir with a Transformer encoder for classification of structured, multi-indicator assessment data: a fixed-order vector of complementary indicators per assessment instance rather than a repeated-measures time series. The reservoir acts as a fixed nonlinear feature map over the indicator vector, while self-attention, made position-aware over the fixed column order, learns how each indicator’s evidence contributes to the final decision, so the two components, together, capture local, indicator-level detail and global, cross-indicator interactions within a single, end-to-end trainable model. Interpretability is treated as a first-class design requirement rather than an afterthought: the architecture is paired with an explainability layer combining SHAP feature attribution (reported both globally and per class), the model’s own attention weights, a deletion/insertion faithfulness test that quantitatively verifies which inputs the model actually relies on, and counterfactual maps that translate a prediction into an actionable, inspectable recommendation. We evaluate the architecture on a concrete case study, classifying Braille-literacy instructional recommendations from 15 pedagogical indicators grouped into three categories (Mangold’s, ABKL, and Progresar), using a benchmark of 900 real assessment instances (630 used, together with a class-conditional augmentation procedure, to build a 2100-instance training set) with validation and test partitions (135 instances each) kept exclusively real. On this benchmark, the tuned model reached 85.33% accuracy, 85.90% macro-precision, 85.33% macro-recall, an F1 score of 85.25%, and an AUC of 0.95 on the real test set. SHAP attribution, attention weights, and the faithfulness test converge on the same two dominant indicators (response time and error count): removing them alone collapses accuracy to chance, while retaining only them recovers most of the model’s accuracy. We report this transparently alongside a comparison against ESN-only, Transformer-only, and tabular baselines (logistic regression, decision tree, random forest, XGBoost, and an MLP) on the same data and discuss what the hybrid architecture and its explainability pipeline add beyond what the two dominant indicators already explain and how the approach generalizes to other tabular and mixed-granularity assessment settings that require both predictive accuracy and a verifiable account of what drove each decision. Full article