Search Results (472)

Small-scale and urban indoor farms typically rely on manual observation, which delays stress detection and yields inconsistent crop quality. While large language models (LLMs) and retrieval-augmented generation (RAG) have been explored for agricultural advisory systems, their integration into a single cloud-free indoor-farming platform that couples multimodal symptom interpretation with autonomous environmental control remains largely unexamined. This study presents an integrated platform built around an agentic AI advisory pipeline that runs entirely on-device on commodity hardware. The pipeline couples a RAG-grounded Mistral 7B language model with a LLaVA 7B vision-language model through condition-based routing, intent classification, multi-step reasoning, and an LLM validation gate, delivering context-aware text and image-based symptom-level guidance from a conversational interface. The advisory layer operates alongside vision-based plant monitoring and a deliberately isolated threshold-based control layer, in which an ESP32 microcontroller autonomously actuates irrigation and lighting against predefined thresholds while a Raspberry Pi 5 performs continuous plant detection and browning monitoring. On Cos lettuce, the advisory pipeline achieved 82.00% weighted accuracy across 50 queries spanning health, symptom, watering, pest, root-health, and growth-stage categories, scored against established plant pathology and postharvest references, with no incorrect responses recorded. The study contributes the design of an agentic advisory pipeline and its integration into a working, cloud-free indoor-farming platform, providing an on-device foundation for intelligent small-scale farming. Full article

Cross-border trade security risks are often embedded in inconsistencies among trade documents, logistics trajectories, hardware sensor states, and financial settlement activities. Existing methods primarily rely on structured declaration fields, making it difficult to verify digital declarations against actual physical processes or to generate complete evidence suitable for regulatory review. To address these challenges, TradeSense-EQA is proposed as a cross-border trade security anomaly detection and evidence-grounded English question-answering framework. Multisource sensing information, including trade documents, GPS/AIS trajectories, RFID records, electronic seal events, port weighing data, temperature and humidity measurements, vibration signals, container door states, and visual images, is jointly modeled within the framework. The reliability-aware representation module dynamically adjusts sensing-channel weights according to data missingness, sampling intervals, device health states, and communication quality. The trade-process-constrained module identifies anomalies across declaration, packing, transportation, transshipment, arrival, and customs clearance stages and generates process-consistent evidence chains. The evidence-grounded question-answering module answers English trade risk questions on the basis of verified documentary fields and sensor records, while confidence estimation and abstention mechanisms are incorporated to reduce factual hallucinations. Experimental results demonstrate that TradeSense-EQA achieved an Accuracy of $0.918$, a Precision of $0.909$, a Recall of $0.897$, a Macro-F1 of $0.903$, and a ROC-AUC of $0.958$ on the cross-border trade anomaly detection task, outperforming baseline methods including XGBoost, LightGBM, TCN, Transformer, BERT, CLIP, and VisualBERT. On the English trade risk question-answering task, Exact Match, Token-level F1, BLEU, ROUGE-L, and BERTScore reached $0.782$, $0.851$, $0.668$, $0.801$, and $0.934$, respectively. Ablation results further confirmed the effectiveness of hardware sensing input, reliability-aware weighting, declaration–fact alignment, process-graph reasoning, and evidence-constrained generation. The proposed framework provides a reliable, interpretable, and auditable artificial intelligence-driven sensing solution for customs supervision, port security, international logistics review, and trade-background investigation. Full article

Oxygen-related absorption features provide fundamental constraints for passive atmospheric remote sensing in the reflected-solar spectrum. Because molecular oxygen is well-mixed in the dry atmosphere, O₂ absorption links measured radiance to atmospheric mass, pressure, and effective photon path length, while O₂-O₂ (O₄) collision-induced absorption provides complementary sensitivity to lower-tropospheric photon paths. This review synthesizes the spectroscopic basis, radiative-transfer mechanisms, satellite implementations, retrieval algorithms, and atmospheric applications of O₂ and O₄ measurements from the ultraviolet to the shortwave infrared. Particular emphasis is placed on the O₂ B-band near 687 nm, the O₂ A-band near 760 nm, O₄ bands in the UV–visible range, and the O₂ band near 1.27 µm. These features support retrievals of cloud fraction, cloud pressure, optical centroid pressure, aerosol layer height, surface pressure, dry-air column abundance, and light-path corrections for greenhouse gas observations. We review major algorithmic approaches, including cloud-as-reflecting-boundary models, cloud-as-layer models, DOAS-based retrievals, optimal-estimation frameworks, photon path-length distribution methods, and machine learning or hybrid techniques. Key applications include cloud climatology, aerosol vertical characterization, air mass factor correction, XCO₂ and XCH₄ retrievals, carbon-cycle studies, and multi-mission data integration. Remaining challenges include spectroscopic uncertainty, aerosol and cloud scattering degeneracy, surface bidirectional reflectance, three-dimensional radiative-transfer effects, wavelength-dependent path mismatch, and inconsistent uncertainty characterization. Future progress will depend on improved spectroscopy, active–passive validation, multi-angle polarimetry, physically constrained machine learning, and harmonized multi-mission retrieval frameworks. Full article

Background: Multiple chemical sensitivity (MCS), a recognized disability, is a chronic, multisystem condition in which exposure to common chemical substances provokes adverse health effects and functional impairment. Although population-level data and emerging clinical research support the pathophysiological basis of MCS, individuals living with the condition frequently encounter barriers to recognition and accommodation across employment, healthcare, and housing systems. This study examines patterns of accommodation requests, refusals, and associated socioeconomic impacts among adults living with MCS in Canada, with particular attention to how institutional recognition shapes access to supports. To our knowledge, this is the first national Canadian study to examine accommodation request outcomes across both employment and housing among adults with MCS, combining quantitative and qualitative data. A national cross-sectional survey of 119 Canadian adults living with MCS was conducted between January and February 2021. Quantitative data were analyzed descriptively to assess accommodation requests, outcomes, employment status, income, and housing stability. Qualitative responses were analyzed thematically to contextualize participants’ experiences of disclosure, and accommodation outcomes. Participants were predominantly female (87%) with post-secondary education (90%). Among participants who reported being unemployed (29%), 94% attributed it primarily to MCS-related limitations. Accommodation denial was common: 85% of respondents reported requesting accommodations, and 78% of those experienced refusals. On housing conditions, 52% reported living in unsafe housing (defined as housing with the presence of mould or symptom-triggering exposures), while 17% reported being homeless (e.g., moving frequently, personal vehicle, tent). Findings should be interpreted in light of the cross-sectional design, self-reported data, and community-based recruitment, which may limit generalizability. The findings suggest that inconsistent institutional recognition of MCS can function as a barrier to accommodation, contributing to health, economic, and social inequities. Addressing these barriers requires clarified guidance on the duty to accommodate multiple chemical sensitivity, proactive scent-free and lowest-emission policies in workplaces and housing, and improved healthcare provider education to support accommodation documentation. Full article

Maintaining durable road infrastructure is crucial for reducing resource consumption, minimizing repair costs, and supporting sustainable urban mobility. However, accurately detecting small and morphologically diverse pavement cracks remains challenging due to variations in lighting, road textures, and crack shapes across different geographic regions. YOLO (You Only Look Once) is one of the most widely adopted deep learning (DL) frameworks for object detection. Traditional inspection methods are labor-intensive and often inconsistent, while existing DL models can be computationally heavy or limited to single crack types, restricting real-time deployment and scalability. To address these challenges, this study presents YOLO26s, a lightweight DL model for multi-class pavement crack detection across diverse environmental and geographic conditions. Using a curated subset of 6972 annotated images from the Road Damage Dataset 2022, YOLO26s identifies four crack types: longitudinal, transverse, pothole, and alligator cracks. Compared to baseline models (YOLOv8s, YOLOv8n, YOLO26n), YOLO26s achieves higher detection accuracy (mAP@0.5 = 89.0%) while reducing computational complexity by 14.3% in parameters and 7.7% in FLOPs, enabling real-time deployment on edge devices. By facilitating early and accurate crack detection, the proposed approach supports proactive maintenance, extends pavement lifespan, and reduces material and energy usage, contributing to more sustainable road network management. These findings highlight the potential of efficient AI-driven inspection systems to enhance environmental and economic sustainability in civil infrastructure. Full article

Background: Amyotrophic lateral sclerosis (ALS), the most common type of motor neurone disease (MND), is a devastating diagnosis that often leads to mortality within 2–5 years of symptom onset. Respiratory failure and aspiration pneumonia both associated with respiratory muscle weakness are the most common causes of death. Difficult to diagnose and devastating in its prognosis, much research has aimed to identify a reliable biomarker to diagnose ALS, prognosticate and improve enrolment into clinical trials to further research efforts. Over the last few decades, neurofilaments (NFs) have emerged as promising biomarkers, especially neurofilament light chain (NFL) and phosphorylated neurofilament heavy chain (pNFH). This review aims to summarise the current evidence for use of NFs as biomarkers in ALS. Current Evidence: Higher levels of NFL and pNFH are measured in CSF than in serum, and levels in CSF and serum are correlated. High CSF NFL, serum NFL and CSF pNFH levels could differentiate patients with ALS from healthy controls, other neurological disease, neurodegenerative controls (without MND), other MND subtypes and ALS disease mimics; however, studies reported a high degree of heterogeneity irrespective of which media or NFs have been used. The number of studies examining NFs to predict respiratory failure in patients with ALS is low. Conclusions and Future Directions: Despite numerous studies consistently reporting higher NF levels in ALS compared to various controls, their clinical value is limited due to high heterogeneity of the results and inconsistencies in proving its prognostic value. Further understanding the relationship between NF levels and respiratory failure is paramount to improve the quality of life of patients with ALS and increase survival. Full article

Artificial light at night (ALAN) is an increasing component of urban environmental change, affecting vegetation dynamics and ecosystem functioning. Satellite nighttime light (NTL) data serve as proxies for urbanization and artificial illumination, aiding the analysis of vegetation responses to human pressures. However, NTL–vegetation relationships are often poorly synthesized, and ALAN is rarely included in frameworks linking urban vegetation, climate, and human drivers. Drawing on a 2014–2025 Scopus and Web of Science search, this review of 22 articles categorizes findings as (i) Lights Track Urbanization, (ii) Vegetation Modulates Light, and (iii) ALAN Shapes Ecology. Results show strong geographical concentration in China, followed by the United States, and high heterogeneity in sensors, metrics, and methods. Increasing nighttime radiance is consistently associated with vegetation decline and higher environmental pressure, while vegetation modulates light through canopy structure and phenology. ALAN effects on plant phenology are reported but vary relative to climatic drivers and are highly context-dependent. Despite these advances, the field remains methodologically inconsistent and geographically biased. This review highlights the need for harmonized multi-sensor frameworks that integrate radiance, vegetation, and climate data to improve assessments of urban environmental change and to support biodiversity conservation and light-sensitive urban planning, thereby preserving ecosystem service functions. Full article

Change detection performance in remote sensing is highly sensitive to illumination variations between bi-temporal images. In real-world scenarios, low-light conditions and inconsistent brightness often lead to degraded feature representations and unreliable detection results. To address this issue, this paper proposes a Consistent Photometric Enhancement Network (CPEN) for change detection under complex illumination conditions. Unlike conventional low-light enhancement methods applied independently to each image, CPEN explicitly enforces photometric consistency between bi-temporal images before and during an enhancement procedure. The proposed framework applies a low-light compensation mechanism to reduce photometric discrepancies between image pairs, which is followed by a bi-temporal enhancement module that jointly improves brightness while preserving shared structural information. Using the photometrically consistent and enhanced images, a change detection module is employed to extract reliable features and generate accurate change maps. Experimental results show that CPEN achieves F1/IoU scores of 90.39/82.67, 80.71/70.58, 78.34/64.26, and 86.26/79.02 on LEVIR-CD, PRCV-CD, SYSU-CD, and the real-world NPULL-CD dataset, respectively, demonstrating its robustness under both simulated and real illumination variations. Full article

Forests play an important role in timber and fiber production, carbon storage, biodiversity conservation, and various other ecosystem services, necessitating accurate and scalable inventory methods. Recent advances in remote sensing have enabled large-scale forest monitoring; however, challenges remain in extracting reliable information across varying spatial, temporal, and environmental conditions. Deep learning has emerged as a promising tool for addressing these limitations by learning complex patterns from diverse remote sensing data sources. This review synthesizes deep learning applications in forest inventory methods across three tasks: tree counting and localization, tree species identification, and tree measurement. In total, we evaluated 122 unique primary studies (37 for tree counting and localization, 57 for species identification, and 29 for tree measurement, with one study contributing to both the counting/localization and measurement tasks) spanning terrestrial, unmanned aerial vehicle (UAV), airborne, and satellite platforms, with a primary focus on optical imagery, Light Detection and Ranging (LiDAR) data, and their fusion. Across these studies, deep learning models frequently outperformed conventional machine learning and statistical baselines, with reported gains including up to 18% improvements in biomass estimation accuracy from data fusion and individual-tree species classification accuracies exceeding 90% for select architectures. However, performance differences were influenced strongly by forest structure, species complexity, sensor capability, and validation design. Counting and localization were generally more reliable in plantations than in complex natural or urban forests, while LiDAR was particularly valuable in dense, multilayer canopies. Species-identification accuracy was highest in studies with small, distinctive species sets, whereas mixed stands with many species showed lower accuracy. Only about a third of the reviewed studies (42 of 122) were externally validated on data or sites independent of model training, and reference data for tree measurement tasks were rarely based on direct destructive sampling. External validation often revealed lower performance than within-study testing, suggesting that reported accuracies may overestimate performance in new locations or conditions. Major advances are evident in the growing use of high-resolution UAV and smartphone-based imagery for tree-level analysis, the continued value of LiDAR for structural characterization, and the increasing integration of multimodal data fusion to improve detection, classification, and measurement accuracy. Persistent challenges include the limited availability of high-quality reference data, class imbalance and inconsistent species coverage, and weak model transferability across forest types, environmental conditions, and geographic regions. Future progress will likely depend on three priorities: development of larger and more standardized labeled datasets, stronger integration of structural, spectral, and phenological information, and the design of more transferable and application-oriented deep learning frameworks. Overall, this review provides a comprehensive, quantitatively grounded overview of deep learning-driven forest inventory methods and outlines future directions for improving scalability and applicability in forest monitoring and management. Full article

Built and ambient features of hospital ward design may influence patient safety and inform future hospital construction and refurbishment. However, evidence on their overall impact remains fragmented. This review synthesised available evidence and highlighted key differences from previous reviews. A comprehensive search identified studies published between 1993 and 2023 involving hospital inpatients. Data were extracted from eligible systematic reviews and subsequently from cited primary studies to synthesise findings and address inconsistencies. Meta-analyses and effect size calculations were performed where appropriate. A total of 31 reviews and 50 primary studies were included. Fall outcomes mainly involved geriatric patients, while infection and delirium outcomes were derived from mixed adult populations. Compared with multi-bedrooms, single-patient rooms were associated with increased fall rates and numbers of fallers, but a lower proportion of patients and rates of hospital-acquired infection and delirium. Natural light and vinyl flooring showed non-significant trends toward reduced delirium and fewer fallers, respectively. Vinyl flooring was associated with higher injurious fall rates. Certainty of evidence was low to very low. Hospital ward design may substantially affect patient safety and quality of care, particularly for older adults. However, effects may be favourable or unfavourable depending on the outcome considered. Further empirical and economic research is needed to balance these competing outcomes. Full article

Three-dimensional concrete printing (3DCP) is an emerging additive manufacturing technology that has attracted significant international interest as a potential alternative to conventional construction methods. Existing research has largely focused on isolated technical domains, resulting in fragmented assessments of the technology and inconsistencies between the construction systems evaluated. This review addresses that gap through a holistic review from technical, architectural, environmental, and economic perspectives, with particular emphasis on its applicability within the New Zealand context. The review found that the multifunctional benefits of 3DCP often cause interdependencies between performance domains, which creates evaluation challenges for individual discipline assessments. Several key barriers to adoption within New Zealand were identified, such as limited evidence of the seismic performance of 3DCP structures, a lack of specific regulatory acceptance pathways, and economic scalability. Nevertheless, 3DCP construction was found to demonstrate suitable technical performance to comply with the New Zealand Building Code (NZBC), though this often involved supplementary construction that failed to fully capitalise on the benefits of 3DCP. Economic, regulatory, and social pressures may be driving re-standardisation in built residential projects, reducing the technology’s architectural differentiation from conventional construction and relinquishing one of the technology’s principal value propositions. Life-cycle assessments suggest environmental performance may be comparable with conventional light timber framing (LTF) where low-carbon strategies are adopted, though this should be re-evaluated as seismically resilient wall typologies are developed. Economic viability remains strongly dependent on deployment scale, with breakeven occurring at approximately eleven dwellings, suggesting the technology will be best suited for large-scale developments and prefabrication facilities. Full article

Accurate 24 h photovoltaic (PV) power forecasting is essential for day-ahead scheduling, storage operation, reserve planning, and market participation. However, published deep learning comparisons are often difficult to reproduce and interpret because they use inconsistent datasets, forecasting horizons, baselines, evaluation metrics, and leakage-control procedures. From a software engineering perspective, this limits the trustworthiness, comparability, and practical adoption of AI-based forecasting systems. This paper presents a controlled and reproducible benchmarking framework for evaluating AI-driven forecasting software. The framework is applied to nine deep learning architectures, three non-deep learning reference models, and two persistence baselines for hourly PV-power forecasting at a 350 kWp rooftop installation near Edinburgh, Scotland. All models were evaluated under a consistent experimental protocol, including the same chronological train–validation–test split, a 32-feature meteorological and solar-geometry input set, a 24-step forecasting horizon, capacity-normalised mean absolute error (NMAE), and Bayesian hyperparameter optimisation. The results show that TCN-LSTM achieved the best aggregate H24 performance with 7.22% NMAE, narrowly outperforming CPWformer-DEC at 7.28% and CT-PatchTST at 7.31%. LightGBM ranked fourth at 7.35% with fixed hyperparameters, outperforming six of the nine deep learning models. The top three models differed by only 0.09 percentage points, indicating that architectural superiority cannot be established reliably without significance testing and operational diagnostics. Per-horizon analysis showed that CT-PatchTST and S-Mamba performed best at the nearest forecast steps, whereas TCN-LSTM provided the most stable far-horizon profile. Peak-power diagnostics further revealed that aggregate NMAE can mask operational shortcomings, as Naive Persistence outperformed all deep learning models in high-output peak detection. The findings highlight the importance of reproducible benchmarking, leakage safeguards, horizon-aware evaluation, and operationally meaningful diagnostics in trustworthy AI software evaluation. The novelty of this work lies not in proposing a new architecture but in a controlled, reproducible framework that benchmarks fourteen forecasters under identical conditions, with explicit leakage safeguards, per-horizon reporting, and operationally meaningful peak diagnostics, enabling claims of architectural superiority to be made trustworthy rather than merely favourable. Architecture selection for PV forecasting should therefore consider not only aggregate accuracy but also reliability, interpretability of evaluation outcomes, and deployment-relevant performance behaviour. Full article

LiDAR sensors provide reliable physical distance measurements using laser signals, enabling accurate acquisition of 3D information for various optical systems. However, they are costly, require significant weight and space, and their reliability and accuracy degrade under adverse environmental and weather conditions. In contrast, thermal cameras operating in the infrared spectrum can capture stable visual information even in challenging scenarios such as nighttime, low-light, and rain. However, they cannot directly provide the physical 3D depth information that LiDAR offers. To design efficient optical systems, there is a growing need for techniques that transform thermal image data into LiDAR-like depth information. While deep learning models can theoretically learn direct mappings between thermal and LiDAR modalities, the scarcity of acquiring paired thermal–LiDAR datasets and the difficulty of acquiring them make this task challenging. In this paper, we propose a thermal image-to-LiDAR depth transformation framework. Our method leverages large-scale pretrained visual models for depth estimation to generate initial depth predictions from thermal inputs. Since pretrained RGB-based models face a modality gap when applied to thermal data, we introduce a two-stage depth refinement. Stage 1 corrects global scale inconsistencies, and Stage 2 refines local structural details. Experiments on the MS2 dataset demonstrate that the proposed framework consistently improves the initial DepthPro outputs across day, night, and rainy conditions. Both quantitative metrics and qualitative comparisons show that RGB-pretrained depth predictions can provide useful structural cues for thermal depth estimation when their global scale and local structural errors are explicitly refined. Full article

Distinguishing seizures from other causes of transient loss of consciousness in the emergency department (ED) is challenging. This PRISMA-guided systematic review evaluated serum brain injury biomarkers for the acute diagnosis of seizures in adults. We searched PubMed and Web of Science for studies published between 2015 and 2025 and included 14 studies in which blood sampling occurred shortly after the event, reflecting the ED diagnostic window. Given the heterogeneity across studies, the overall certainty of the evidence was low. Neuron-specific enolase (NSE) and ubiquitin carboxyl-terminal hydrolase L1 (UCH-L1) were consistently elevated after epileptic seizures compared to healthy controls. NSE effectively differentiated seizures from syncope, while UCH-L1 and glial fibrillary acidic protein (GFAP) distinguished epileptic from psychogenic non-epileptic seizures (PNESs). Neurofilament light chain (NfL) remained stable after a single seizure but increased markedly in status epilepticus (SE). S100B and BDNF results were inconsistent. Although no single biomarker serves as a standalone test, NSE and UCH-L1 are promising complementary diagnostic tools for identifying epileptic seizures in adults. Furthermore, NfL is a strong candidate marker for SE, reflecting neuroaxonal injury. Larger prospective, standardized studies are needed before routine ED implementation. Full article

Deploying retired lithium-ion batteries in photovoltaic energy storage systems is a promising second-life application, but heterogeneous aging and internal inconsistencies can induce the barrel effect, reducing available capacity and accelerating pack degradation. Existing equalization methods mainly rely on internal battery states and often neglect external irradiance fluctuations. To address this issue, this study proposes an irradiance-aware adaptive fuzzy active equalization strategy based on a multichannel bidirectional flyback converter. A second-order RC equivalent circuit model with a fifth-order OCV–SOC mapping is established to describe the dynamic behavior of retired cells. Then, solar irradiance and its rate of change are introduced into a dual-input fuzzy controller to adaptively regulate the equalization duty cycle according to both SOC inconsistency and PV input fluctuation. A saturation function constrains the active duty cycle below 0.5 to maintain discontinuous conduction mode operation and avoid transformer core saturation. Simulation results under rapid cloud occlusion, stable high irradiance, and persistent weak light show that the proposed strategy reduces equalization time by 13.8%, 4.4%, and 8.4%, respectively, compared with SOC-only fuzzy control. Under a publicly measured irradiance condition, the proposed strategy achieves the shortest equalization time of 3267.4 s, reducing the time by 24.2%, 27.7%, 29.0%, and 32.9% compared with traditional threshold-based, SOC-only fuzzy, maximum–minimum SOC, and PID-based strategies, respectively. Full article