Search Results (24,893)

Background/Objectives: D-dimer, a fibrin degradation product, is associated with tumor growth and metastasis. In breast cancer, high concentrations of D-dimer are linked to more advanced disease stages and metastatic spread. This research aimed to examine the relevance of D-dimer levels in estrogen and progesterone hormone receptor (HR)-positive breast cancer. Methods: This retrospective single-center cohort study included patients with HR-positive breast carcinoma who underwent adjuvant or palliative radiotherapy in Türkiye. Pre- and post-radiotherapy blood test results, including D-dimer levels, were required. D-dimer, lymphocyte percentage, and interleukin-6 levels were measured for evaluation. All statistical analyses were performed using R software (version 4.4.2) to evaluate associations between D-dimer levels and other laboratory parameters. Univariate and multivariate Cox proportional hazards regression were performed to identify prognostic factors for progression-free survival (PFS) and overall survival (OS). Statistical significance was defined as p < 0.05. Results: Elevated D-dimer levels were associated with worse Eastern Cooperative Oncology Group performance status, advanced disease stages, metastasis, elevated IL-6 and CRP levels, and lower lymphocyte counts. Pre-RT D-dimer was a strong prognostic factor. Patients with D-dimer ≤ 0.3 µg/mL showed significantly superior OS and PFS (>60 months; p < 0.001), with only one event, and this remained significant in multivariate analysis (OS: HR 4.55, 95% CI 1.89–11.3; p = 0.002; PFS: HR 3.43, 95% CI 1.54–7.8; p = 0.004). Similarly, D-dimer ≤ 0.5 µg/mL was associated with improved OS (4/72 vs. 19/40 events; p < 0.001) and longer PFS, confirmed in multivariate analysis (OS: HR 4.37, 95% CI 1.72–9.86; p = 0.002; PFS: HR 3.88, 95% CI 1.67–9.1; p = 0.003), whereas levels > 0.5 µg/mL predicted worse outcomes. Using a 0.65 µg/mL cutoff, patients with D-dimer > 0.65 µg/mL had significantly shorter OS (median 25.5 months; 95% CI, 18–NA) compared with those ≤0.65 µg/mL (median not reached; p < 0.001), and this remained independently significant (OS: HR 5.10, 95% CI 1.9–13.6; p < 0.001; PFS: HR 4.68, 95% CI 1.83–11.9; p = 0.002). Conclusions: D-dimer is an accessible, non-invasive biomarker with predictive and prognostic significance in HR-positive breast cancer. Elevated D-dimer levels are suggestive of a more aggressive disease and poorer survival outcomes. This has the potential to facilitate early assessment of treatment efficacy and disease progression. This study has several limitations. Its retrospective, single-center design may introduce selection bias and limit generalizability. Although the sample size was sufficient to detect significant associations, validation in larger, multi-center cohorts is warranted to confirm the prognostic value of D-dimer. Full article

Background: Sarcopenia is a progressive muscle disorder associated with metabolic syndrome (MS), in which early impairments in muscle strength and quality precede muscle mass loss. Simple, non-invasive measures such as handgrip strength, estimated appendicular skeletal muscle mass (eASM), and phase angle (PA) may aid early detection, while adipokines link muscle dysfunction to metabolic regulation. Objective: In the present study, we aimed to evaluate the association between sarcopenia markers and PA in patients with MS. Methods: A cross-sectional study was conducted in patients with MS, at a third-level hospital in Mexico City. Sarcopenia was assessed by handgrip strength and eASM; body composition and PA were measured using bioelectrical impedance; and plasma adipokines were quantified by ELISA. Results: Seventy-four (mean age, 57.7 years; 75% female; BMI, 32.5 kg/m²) participants with MS were included. Handgrip strength correlated with eASM (r = 0.64; p < 0.01) and PA (rho = 0.43; p < 0.01), and eASM also correlated with PA (rho = 0.40; p < 0.01) and predicted higher PA values (OR = 2.74; p = 0.042). The sarcopenic subgroup had lower brachial circumference and plasma adiponectin. Conclusions: Sarcopenia is frequent in MS and associated with lower adiponectin, suggesting a vulnerable condition. Functional/structural markers of sarcopenia showed significant correlation with PA, whereas combined methods may enhance the early detection and management of muscle deterioration in metabolic disease. Full article

Background/Objectives: Ampullary adenomas are neoplasms of the gastrointestinal tract with malignant potential. They are thought to develop through pathways similar to those involved in colorectal neoplasia. This study aimed to determine the prevalence of colonic polyps in patients with ampullary adenoma. Methods: This retrospective study included a total of 35 patients with ampullary adenoma diagnosed between 2023 and 2024 and 105 sex-matched controls. Colonoscopic findings of the patient and control groups were compared with respect to polyp prevalence. In addition, the effects of dysplasia grade of the ampullary adenoma and patient age on polyp prevalence were evaluated. Results: The study included 35 patients (57% male) and 105 controls (59% male). The mean age was 67.06 ± 13.32 years in patients and 61.28 ± 8.42 years in controls. Colonic polyps were detected in 13 (57%) patients in the low-grade dysplasia (LGD) group, 6 (66%) patients in the high grade dysplasia (HGD) or adenocarcinoma group, and 54 (51%) patients in the control group (p = 0.02). After adjusting for age, colonic polyps remained significantly more frequent in the adenoma group than in controls (p = 0.05). Polyp prevalence was not associated with dysplasia grade on ampullary biopsy, and no significant differences were observed between groups regarding polyp histopathology, location, or size. Conclusions: In conclusion, our study indicates that colorectal polyp prevalence is increased among patients with ampullary adenomas and that this association may be independent of age as well as dysplasia severity. Therefore, colonoscopic evaluation may be recommended for all patients diagnosed with ampullary adenoma. Full article

Background/Objectives: Identification of tools to minimize opioid-related harms is critical in the U.S. The purpose of this study was to better understand community responder and healthcare provider perceptions and preferences regarding the design and function of a potential new opioid sensor device (OSD). Methods: Adults aged ≥ 18 years employed as community responders or healthcare providers in Alabama were recruited via email to participate in an anonymous online cross-sectional survey informed by the Unified Theory of Acceptance and Use of Technology (UTAUT). Primary outcomes were assessed via multiple-choice and 7-point Likert-type scales (1 = strongly disagree, 7 = strongly agree) and included the following topics: (1) past OSD utilization (4 items); (2) perceived importance of OSD design elements (15 items); (3) OSD function and cost preferences (3 items); and (4) UTAUT measures including perceived usefulness of OSDs (3 items), ease of use (4 items), social factors (4 items), resources (4 items), concerns (3 items), and intentions (3 items). Differences in UTAUT measures across professions were assessed via Mann–Whitney U tests, and predictors of OSD utilization intention were analyzed via multiple linear regression. Results: Respondents (N = 145) included pharmacists (40.0%), nurses (23.4%), physicians (14.5%), behavioral health (4.8%), social work (4.8%), and law enforcement (0.7%). Availability in hospital emergency departments was rated as the most important device element (mean [SD] score: 6.66 [0.80]), followed by sensitivity and specificity of the test (6.42 [0.98]), rapid detection time (6.42 [0.88]), ability to detect opioids in a broad range of substance (6.42 [0.93]), and availability in law enforcement offices (6.33 [1.08]). A 2–5 min detection time was rated as reasonable by 32.6% of respondents, with 53.0% preferring to pay <USD 15 per test. There were no statistically significant differences in UTAUT scale scores across professions. Perceived usefulness (β = 0.493; p < 0.001), social acceptance (β = 0.281; p = 0.023), and resource availability (β = 0.708; p = 0.002) were positive predictors and perceived ease of use was a negative predictor (β = −0.472; p = 0.007) of intention to use an OSD. Conclusions: Newly developed OSDs should consider prioritizing accessibility in hospital emergency departments and law enforcement offices, ability to detect a broad range of opioids, detection time between 2 and 5 min, and cost less than USD 15 per test. Future research may explore perspectives from a more diverse sample across multiple states and different professional roles. Full article

Stance detection has emerged as an essential tool in natural language processing for understanding how individuals express agreement, disagreement, or neutrality toward specific targets in social and online discourse. It plays a crucial role in bilingual and multilingual environments, including English-Arabic social media ecosystems, where differences in language structure, discourse style, and data availability pose significant challenges for reliable stance modelling. Existing approaches often struggle with target awareness, cross-lingual generalization, robustness to noisy user-generated text, and the interpretability of model decisions. This study aims to build a reliable, explainable target-aware bilingual stance-detection framework that generalizes across heterogeneous stance formats and languages without retraining on a dataset specific to the target language. Thus, a unified dual-encoder architecture based on mDeBERTa-v3 is proposed. Cross-language contrastive learning offers an auxiliary training objective to align English and Arabic stance representations in a common semantic space. Robustness-oriented regularization is used to mitigate the effects of informal language, vocabulary variation, and adversarial noise. To promote transparency and trustworthiness, the framework incorporates token-level rationale extraction, enables fine-grained interpretability, and supports analysis of hallucination. The proposed model is tested on a combined bilingual test set and two structurally distinct zero-shot benchmarks: MT-CSD and AraStance. Experimental results show consistent performance, with accuracies of 85.0% and 86.8% and F1-scores of 84.7% and 86.8% on the zero-shot benchmarks, confirming stable performance and realistic generalization. Ultimately, these findings reveal that effective bilingual stance detection can be achieved via explicit target conditioning, cross-lingual alignment, and explainability-driven design. Full article

With the continuous advancement of the “dual-carbon” strategy, the penetration of renewable energy sources such as wind and photovoltaic (PV) power has steadily increased, imposing more stringent requirements on the safe and stable operation of modern power systems. As the core components of these systems, critical electrical devices operate under harsh conditions characterized by high voltage, strong electromagnetic interference (EMI), and confined high-temperature environments. Their operating status directly affects the reliability of the power supply, and any fault may trigger cascading failures, resulting in significant economic losses. To address the issues of low inspection efficiency, limited fault-identification accuracy, and unstable data transmission in strong-EMI environments, this study proposes an intelligent inspection system for power equipment based on multi-technology integration. The system incorporates a redundant dual-mode wireless transmission architecture combining Wireless Fidelity (Wi-Fi) and Fourth Generation (4G) cellular communication, ensuring reliable data transfer through adaptive link switching and anti-interference optimization. A You Only Look Once version 8 (YOLOv8) object-detection algorithm integrated with Open Source Computer Vision (OpenCV) techniques enables precise visual fault identification. Furthermore, a multi-source data-fusion strategy enhances diagnostic accuracy, while a dedicated monitoring scheme is developed for the water-cooling subsystem to simultaneously assess cooling performance and fault conditions. Experimental validation demonstrates that the proposed system achieves a fault-diagnosis accuracy exceeding 95.5%, effectively meeting the requirements of intelligent inspection in modern power systems and providing robust technical support for the operation and maintenance of critical electrical equipment. Full article

We aim to address the insufficient robustness of navigational line detection for rapeseed seed production sires in complex field scenarios and the challenges faced by existing models in balancing precision, real-time performance, and resource consumption. Taking YOLOv8n-seg as the baseline, we first introduced the ADown module to mitigate feature subsampling information loss and enhance computational efficiency. Subsequently, the DySample module was employed to strengthen target feature representation and improve object discrimination in complex scenarios. Finally, the c2f module was replaced with c2f_FB to optimise feature fusion and reinforce multi-scale feature integration. Performance was evaluated through comparative experiments, ablation studies, and scenario testing. The model achieves an average precision of 99.2%, mAP50-95 of 84.5%, a frame rate of 90.21 frames per second, and 2.6 million parameters, demonstrating superior segmentation performance in complex scenarios. SegNav-YOLOv8n balances performance and resource requirements, validating the effectiveness of the improvements and providing reliable technical support for navigating agricultural machinery in rapeseed seed production. Full article

Background: Falls are a leading cause of morbidity and mortality among older adults, creating a need for technologies that can automatically detect falls and summon timely assistance. The rapid evolution of sensor technologies and artificial intelligence has led to a proliferation of fall detection systems (FDS). This systematic review synthesizes the recent literature to provide a comprehensive overview of the current technological landscape. Objective: The objective of this review is to systematically analyze and synthesize the evidence from the academic literature on fall detection technologies. The review focuses on three primary areas: the sensor modalities used for data acquisition, the computational models employed for fall classification, and the emerging trend of shifting from reactive detection to proactive fall risk prediction. Methods: A systematic search of electronic databases was conducted for studies published between 2008 and 2025. Following the PRISMA guidelines, 130 studies met the inclusion criteria and were selected for analysis. Information regarding sensor technology, algorithm type, validation methods, and key performance outcomes was extracted and thematically synthesized. Results: The analysis identified three dominant categories of sensor technologies: wearable systems (primarily Inertial Measurement Units), ambient systems (including vision-based, radar, WiFi, and LiDAR), and hybrid systems that fuse multiple data sources. Computationally, the field has shown a progression from threshold-based algorithms to classical machine learning and is now dominated by deep learning architectures, such as Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Transformers. Many studies report high performance, with accuracy, sensitivity, and specificity often exceeding 95%. An important trend is the expansion of research from post-fall detection to proactive fall risk assessment and pre-impact fall prediction, which aim to prevent falls before they cause injury. Conclusions: The technological capabilities for fall detection are well-developed, with deep learning models and a variety of sensor modalities demonstrating high accuracy in controlled settings. However, a critical gap remains; our analysis reveals that 98.5% of studies rely on simulated falls, with only two studies validating against real-world, unanticipated falls in the target demographic. Future research should prioritize real-world validation, address practical implementation challenges such as energy efficiency and user acceptance, and advance the development of integrated, multi-modal systems for effective fall risk management. Full article

Background/Objectives: Human papillomavirus (HPV) is the most common sexually transmitted virus worldwide and is frequently detected in women of reproductive age. In this population, HPV-related diseases and their management may affect reproductive health and pregnancy outcomes. This narrative review summarizes the current evidence on HPV infection and HPV-related diseases in relation to fertility, pregnancy, and neonatal outcomes, and discusses preventive strategies, with a particular focus on HPV vaccination. Methods: An international, multidisciplinary team of clinicians from the European Society of Gynaecological Oncology (ESGO) Prevention Committee reviewed the literature on HPV, HPV-related diseases, HPV vaccination, and reproductive outcomes, without time restrictions, prioritizing studies judged to meaningfully reflect the available evidence. Results: The most consistent evidence linking HPV-related conditions to adverse pregnancy outcomes relates to the treatment of cervical precancer, particularly excisional procedures, which are associated with an increased risk of preterm birth and mid-trimester pregnancy loss. In contrast, evidence that maternal HPV detection alone causes adverse pregnancy or neonatal outcomes remains limited and inconsistent. Data on HPV infection and subfertility are scarce and heterogeneous. Management of HPV-related lesions during pregnancy remains challenging and requires careful balancing of maternal safety with avoidance of unnecessary interventions. HPV DNA has been detected in neonatal samples, but convincing evidence for clinically relevant vertical transmission is lacking. Available data indicate that inadvertent HPV vaccination shortly before or during pregnancy is not associated with adverse pregnancy outcomes. Conclusions: Current evidence suggests that reproductive risks are more strongly associated with the treatment of HPV-related diseases than with HPV infection itself. Preventive strategies—especially HPV vaccination—remain central to reducing HPV-related disease burden. Although HPV vaccines are not routinely recommended during pregnancy, evidence supports the safety of inadvertent exposure around conception or during gestation, while potential long-term benefits of vaccination regarding reproductive health require further study. Full article

Objective: To compare metastatic lesion detection on [68Ga]Ga-DOTA.SA.FAPi and [18F]FDG PET/CT in metastatic breast and lung cancers and to assess the relationship between PET-derived imaging parameters and progression-free survival (PFS). Methods: In this prospective dual-cohort study, 45 patients (23 breast cancer, 22 lung adenocarcinoma) underwent paired [68Ga]Ga-DOTA.SA.FAPi and [18F]FDG PET/CT within four weeks. Semiquantitative (SUVmax, SUVmean) and volumetric (MTV, TLG, STV, TLF) PET parameters were measured. Metastatic detection was compared, and correlations with PFS were assessed. Results: In breast cancer, [18F]FDG demonstrated higher primary tumor uptake, whereas [68Ga]Ga-DOTA.SA.FAPi showed lower background activity, resulting in higher tumor-to-background ratios for brain and bone metastases. Whole-body volumetric indices (wbTLG, wbTLF) showed strong inverse correlations with PFS. In lung adenocarcinoma, volumetric FAPi-derived parameters (wbTLF, wbSTV) demonstrated modest but significant correlations with PFS. [68Ga]Ga-DOTA.SA.FAPi PET/CT detected more brain metastases than [18F]FDG PET/CT in both cohorts (breast: 15/15 vs. 8/15; lung: 14/14 vs. 4/14). Conclusions: [68Ga]Ga-DOTA.SA.FAPi and [18F]FDG PET/CT provide complementary diagnostic and prognostic information. In metastatic breast cancer, FAPi-derived volumetric parameters strongly correlate with PFS and improve detection of brain metastases. In lung adenocarcinoma, [68Ga]Ga-DOTA.SA.FAPi PET/CT offers low background uptake and prognostically relevant stromal metrics. These findings support a potential role for integrating [68Ga]Ga-DOTA.SA.FAPi PET/CT into disease staging, prognostication, and treatment monitoring. This study did not involve prospective assignment to health-related interventions and therefore did not require clinical trial registration. Full article

Recent developments in computer vision have significantly enhanced the automation and objectivity of sports analytics. This paper proposes a novel deep learning-based framework for estimating football possession directly from broadcast video, eliminating the reliance on manual annotations or event-based data that are often labor-intensive, subjective, and temporally coarse. The framework incorporates two structurally improved object detection models: YOLOv8-P2S3A for football detection and YOLOv8-HWD3A for player detection. These models demonstrate superior accuracy compared to baseline detectors, achieving 79.4% and 71.1% validation average precision, respectively, while maintaining low computational latency. Team identification is accomplished through unsupervised DBSCAN clustering on jersey color features, enabling robust and label-free team assignment across diverse match scenarios. Object trajectories are maintained via the Norfair multi-object tracking algorithm, and a temporally aware refinement module ensures accurate estimation of ball possession durations. Extensive experiments were conducted on a dataset comprising 20 full-match Video clips. The proposed system achieved a root mean square error (RMSE) of 4.87 in possession estimation, outperforming all evaluated baselines, including YOLOv10n (RMSE: 5.12) and YOLOv11 (RMSE: 5.17), with a substantial improvement over YOLOv6n (RMSE: 12.73). These results substantiate the effectiveness of the proposed framework in enhancing the precision, efficiency, and automation of football analytics, offering practical value for coaches, analysts, and sports scientists in professional settings. Full article

The feather coverage of a laying hen is an important indicator of both its productivity and welfare. Conventional manual feather scoring procedures are laborious, subjective, and stressful for the hens. Thermography offers a modern alternative to addressing these problems. Thermal cameras capture radiative heat loss, which is comparatively greater Classification from featherless areas. Studies have been conducted to establish a standard temperature range that correlates to specific featherless areas. However, such temperature-based approaches have been inconsistent with each other. In contrast, this study used deep learning techniques to automatically assess dorsal feather scores using thermal images. Thermal images (n = 1575) of the dorsal body of cage-free laying hens with varying degrees of feather damage were captured. Manual feather scoring was performed, classifying the image into a feather score (0–2) according to the increasing severity of feather loss. A total of 1222 images were selected, filtering out images of lower quality. Two types of computer vision models, a classification model and an object detection model, were trained and evaluated. A custom convolutional neural network (CNN) was trained to classify thermal images into feather score categories. Additionally, we trained and optimized You Only Look Once (YOLO) object detection models to detect areas of feather damage and predict the feather score. The CNN model achieved an overall accuracy of 0.81, with high precision for severe feather loss. The YOLO-based object detection model was optimum using YOLO11n, which achieved a precision of 0.81, a recall of 0.73 and a mean average precision (mAP) at 0.5 intersection over union (IoU) of 0.84. Results show the potential of combining thermal imaging with deep learning techniques to perform objective, automatic, and scalable feather scoring procedures. Future studies should focus on data diversity, multiple part scoring, and semantic segmentation for robust performance. Full article

Context: Code recommendation plays an important role in improving programming efficiency and software quality. Existing approaches mainly focus on method- or API-level recommendations, which limits their effectiveness to local code contexts. From a multi-stage recommendation perspective, class-level code recommendation aims to efficiently narrow a large candidate code space while preserving essential structural information. Objective: This paper proposes RioCC, a class-level code recommendation framework that leverages deep forest-based code clone detection to progressively reduce the candidate space and improve recommendation efficiency in large-scale code spaces. Method: RioCC models the recommendation process as a coarse-to-fine candidate reduction procedure. In the coarse-grained stage, a quick search-based filtering module performs rapid candidate screening and initial similarity estimation, effectively pruning irrelevant candidates and narrowing the search space. In the fine-grained stage, a deep forest-based analysis with cascade learning and multi-grained scanning captures context- and structure-aware representations of class-level code fragments, enabling accurate similarity assessment and recommendation. This two-stage design explicitly separates coarse candidate filtering from detailed semantic matching to balance efficiency and accuracy. Results: Experiments on a large-scale dataset containing 192,000 clone pairs from BigCloneBench and a collected code pool show that RioCC consistently outperforms state-of-the-art methods, including CCLearner, Oreo, and RSharer, across four types of code clones, while significantly accelerating the recommendation process with comparable detection accuracy. Conclusions: By explicitly formulating class-level code recommendation as a staged retrieval and refinement problem, RioCC provides an efficient and scalable solution for large-scale code recommendation and demonstrates the practical value of integrating lightweight filtering with deep forest-based learning. Full article

In agriculture, the use of Porous Coordination Polymers (PCPs), also known as Metal–Organic Frameworks (MOFs), has emerged as a promising area of research for biological applications, particularly as long-lasting delivery systems for biostimulant chemical compounds. The objective of this study was to evaluate the effect of different concentrations of the hybrid compound La(NO₃)₃@Zn-MOF and La(NO₃)₃·6H₂O on the in vitro growth of sugarcane cv. Mex 69–290. To assess the effect on sugarcane (Saccharum spp. L.), plantlets were grown in flasks containing Murashige and Skoog (MS) liquid medium without growth regulators. Each treatment consisted of three independent culture flasks, each containing three sugarcane plantlets, and different concentrations (0, 2.5, 5, 10, and 20 mg L⁻¹) of La(NO₃)₃@Zn-MOF and La(NO₃)₃·6H₂O were added separately. After 30 days of culture, various growth variables were evaluated, including explant length, number of leaves, number and length of shoots, fresh matter, dry matter, and chlorophyll content (a, b, and total). The 5 mg L⁻¹ concentration of La(NO₃)₃@Zn-MOF increased the number of shoots and leaves in the sugarcane plantlets, and significant increases in fresh and dry matter were observed, while no statistically significant differences were detected in explant length, shoot length, or chlorophyll a, b, and total chlorophyll. However, inhibitory effects were observed at concentrations of 10 and 20 mg L⁻¹ of La(NO₃)₃@Zn-MOF and La(NO₃)₃·6H₂O, respectively. In conclusion, the hybrid compound La(NO₃)₃@Zn-MOF exhibited biostimulatory effects on sugarcane growth and physiology under in vitro conditions, whereas high concentrations of the lanthanum(III) salt caused toxicity symptoms in the sugarcane plantlets. Full article

Automated insulator inspection systems face critical challenges from small object sizes, complex backgrounds, and vulnerability to adversarial attacks, a security concern largely unaddressed in safety-critical power infrastructure. We introduce Faster-YOLOv12n, integrating a FasterNet backbone with SGC2f attention modules and Wise-ShapeIoU loss for enhanced small defect localization. Our architecture achieves 98.9% mAP@0.5 on the CPLID, improving baseline YOLOv12n by 1.3% in precision (97.8% vs. 96.5%), 4.7% in recall (95.1% vs. 90.4%), and 1.8% in mAP@0.5. Through differential data augmentation, we expand training samples from 678 to 3900 images, achieving balanced class distribution and robust generalization across fog, adverse weather, and complex transmission line backgrounds. Comparative evaluation demonstrates superior performance over RT-DETR, Faster R-CNN, YOLOv7, YOLOv8, and YOLOv9, with per-class analysis revealing 99.8% AP@0.5 for defect detection. We provide the first comprehensive adversarial robustness evaluation for insulator defect detection, systematically assessing FGSM, PGD, and C&W attacks across perturbation budgets. Through adversarial training with mixed-batch strategies, our robust model maintains 93.2% mAP@0.5 under the strongest FGSM attacks ($ϵ$ = 48/255), 94.5% under PGD attacks, and 95.1% under C&W attacks ($\tau$ = 3.0) while preserving 98.9% clean accuracy, demonstrating no trade-off between accuracy and robustness. Grad-CAM visualizations demonstrate that attacks disrupt confidence calibration while preserving spatial attention on defect regions, providing interpretable insights into model decision-making under adversarial conditions and validating learned feature representations for safety-critical smart grid monitoring applications. Full article