Search Results (323)

This paper presents a novel adversarial Convolutional Neural Network (ConvNet) method for generating adversarial perturbations in 3D point clouds, enabling gradient-free robustness evaluation of object detection systems at inference time. Unlike existing iterative gradient methods, our approach embeds the ConvNet directly into the detection pipeline at the voxel feature level. The ConvNet is trained to maximize detection loss while maintaining perturbations within sensor error bounds through multi-component loss constraints (intensity, bias, and imbalance terms). Evaluation on a Sparsely Embedded Convolutional Detection (SECOND) detector with the KITTI dataset shows 8% overall mean Average Precision (mAP) degradation, while CenterPoint on NuScenes exhibits 24% weighted mAP reduction across 10 object classes. Analysis reveals an inverse relationship between object size and adversarial vulnerability: smaller objects (pedestrians: 13%, cyclists: 14%) show higher vulnerability compared to larger vehicles (cars: 0.2%) on KITTI, with similar patterns on NuScenes, where barriers (68%) and pedestrians (32%) are most affected. Despite perturbations remaining within typical sensor error margins (mean L2 norm of 0.09% for KITTI, 0.05% for NuScenes, corresponding to 0.9–2.6 cm at typical urban distances), substantial detection failures occur. The key novelty is training a ConvNet to learn effective adversarial perturbations during a one-time training phase and then using the trained network for gradient-free robustness evaluation during inference, requiring only a forward pass through the ConvNet (1.2–2.0 ms overhead) instead of iterative gradient computation, making continuous vulnerability monitoring practical for autonomous driving safety assessment. Full article

To improve the economy and stability of data center green power direct supply, the capacity configuration optimization of wind–light–load storage based on improved particle swarm optimization (PSO) is conducted. According to wind speed, the Weibull distribution of wind output is established, while the Beta distribution of solar output is established according to light intensity. Furthermore, by conducting the correlation analysis, it is indicated that there is a negative correlation between wind and solar output, which is helpful to optimize the mix of wind and solar output. To minimize the yearly average cost of wind–light–load storage, the capacity configuration optimization model is established, where the constraints include wind and solar output, energy storage capacity, balance between wind and solar output and data center load. To solve the capacity configuration optimization model, the improved PSO is adopted, compared to other optimization algorithms, like differential evolution (DE), genetic algorithm (GA) and grey wolf optimizer (GWO); by adjusting the inertia weight factor dynamically, the improved PSO is more likely to escape the local optimal solution. To validate the feasibility of data center green power direct supply with wind–light–load storage, a case study is conducted. By solving the capacity configuration optimization model of wind–light–load storage with the improved PSO, the balance rate between wind–solar output and data center load is improved by 12.5%, while the rate of abandoned wind and solar output is reduced by 17.5%, which is helpful to improve the economy and stability of data center green power direct supply. Full article

To address the challenge of reduced prediction accuracy caused by capacity regeneration during the use of lithium-ion batteries, this study proposes an RUL (remaining useful life) prediction method based on mode decomposition and an enhanced Informer-LSTM hybrid model. The capacity is selected as the health indicator, and the CEEMDAN (complete ensemble empirical mode decomposition with adaptive noise) algorithm is employed to decompose the capacity sequence into high-frequency and low-frequency components. The high-frequency components are further decomposed and predicted using the Informer model, while the low-frequency components are predicted with an LSTM (long short-term memory) network. Pearson correlation coefficients between each component and the original sequence are calculated to determine fusion weights. The final RUL prediction is obtained through weighted integration of the individual predictions. Experimental validation on publicly available NASA and CALCE (Center for Advanced Life Cycle Engineering) battery datasets demonstrates that the proposed method achieves an average fitting accuracy of approximately 99%, with MAE (mean absolute error) below 0.02. Additionally, both MAPE (mean absolute percentage error) and RMSE (root-mean-square error) remain at low levels, indicating improvements in prediction precision. Full article

This paper presents the design, development, and experimental evaluation of a hybrid wheel–leg guide robot intended to assist blind and elderly people with mobility tasks indoors and outdoors. The design requirements are derived from an analysis of safety, usability, and affordability needs for assisting devices. The resulting design consists of a compact platform with two front leg–wheel assemblies and three additional wheels, two of which are motorized, arranged in a triangular configuration that provides stable support and reliable traction. The proposed locomotion system is innovative because existing guide robots typically rely exclusively on either wheels or legs. In contrast, this hybrid configuration combines the energy efficiency of wheeled locomotion with the capability of leg-assisted stepping, enabling improved terrain adaptability. Experiments with a prototype were carried out in indoor environments, including straight-line motion, turning, and obstacle-overcoming tests. The prototype, with a total weight of 1.9 kg and a material cost of 255 euros, maintained stable movement and achieved a 100% success rate for obstacles up to 30 mm, with partial success up to 40 mm. Additional test results indicate an average cruising speed of 0.1 m/s, and a practical endurance of 4.5–5 h. The proposed design aims to contribute to the development of more inclusive, efficient, and user-centered robotic solutions, promoting greater autonomy and quality of life for blind and elderly people. Full article

Despite rapid proliferation of AI applications in prostate MRI showing impressive technical performance, clinical adoption remains limited. We conducted a comprehensive narrative review of literature from January 2018 to December 2024, examining AI applications in prostate MRI with emphasis on real-world performance and implementation challenges. Among 200+ studies reviewed, AI systems achieve 87% sensitivity and 72% specificity for cancer detection in research settings. However, external validation reveals average performance drops of 12%, with some implementations showing degradation up to 31%. Only 31% of studies follow reporting guidelines, 11% share code, and 4% provide model weights. Seven real-world implementation studies demonstrate integration times of 3–14 months, with one major center terminating deployment due to unacceptable false positive rates. The translation gap between artificial and clinical intelligence remains substantial. Success requires shifting focus from accuracy metrics to patient outcomes, establishing transparent reporting standards, developing realistic economic models, and creating appropriate regulatory frameworks. The field must combine methodological rigor, clinical relevance, and implementation science to realize AI’s transformative potential in prostate cancer care. Full article

Advancement of compact atomic clocks has centered on reducing footprint and power consumption. Such developments come at the cost of the clock’s stability performance. Various commercial and military applications demand reduced size, weight, and power (SWaP) requirements but desire an enhanced stability performance beyond what is achieved with the lower-profile standards, such as Microchip’s chip-scale atomic clock (CSAC) or miniature atomic clock (MAC). Furthermore, a high-performing space-rated clock will enhance small satellite missions by providing capability for alternate PNT, one-way radiometric ranging, and eventual lunar PNT purposes. The MAC is a strong candidate as it has modest SWaP parameters. Enhanced performance improvement to the MAC, particularly in the medium to long-term stability over a day and beyond will strengthen its candidacy as an on-board clock in small satellite missions and other ground-based applications. In this work, using external thermal control methods, we demonstrate an improvement of the MAC performance by at least a factor of five, showing a superior stability of σ_y = 4.2 × 10⁻¹³ compared to the best-performing miniaturized standard on the market for averaging intervals of τ > 10⁴ s up to 4 days. Full article

Standardized driving cycles such as the FTP-75 fail to represent traffic conditions in cities like Guayaquil, where high congestion and varied driving behaviors are not captured by external models. This study aimed to develop representative driving cycles for the city’s most congested urban routes, covering the north, south, center, and west zones. Using the direct method, real-world trips were conducted with an M1-category vehicle equipped with an OBDLINK MX+ device, allowing real-time data collection. Driving data were processed through OBDWIZ software Version 4.30.1 and statistically analyzed using Minitab. From pilot tests, the appropriate sample size was estimated, and normality tests were applied to determine the correct measures of central tendency. The final representative cycles were constructed using a weighting criteria method. The results provided quantified evidence of variations in average speed, idle time, and acceleration patterns across the routes, which were transformed into representative driving cycles. These cycles provide a more accurate basis for emission modeling, vehicle certification, and transport policy design in congested cities such as Guayaquil, and this is the applied impact that is highlighted in our contribution. Furthermore, the developed cycles provide a foundation for future research on emission modeling and the design of sustainable transport strategies in Latin American cities. Full article

Background: Hypotensive events may occur during surgical interventions and are associated with major postoperative complications, depending on their duration and severity. Intraoperative hemodynamic goal-directed therapy can reduce postoperative complications and mortality in high-risk surgeries and high-risk patients. The study hypothesis was that a proactive approach by hypotension predictive index (HPI) is more effective than a reactive goal-directed therapy (GDT) in reducing the number of hypotensive events during radical cystectomy and that this is associated with improved postoperative outcomes. Methods: The study was a single-center prospective observational study conducted at Galliera Hospital, from November 2019 to February 2025, with a before-after population of sixty-seven patients with reactive approach (GDT group) and sixty-five patients with a proactive approach (HPI group) undergoing radical cystectomy, managed with a standardized ERAS protocol and invasive or non-invasive hemodynamic monitoring. The aim of the study was to compare the incidence, duration, and severity of intraoperative hypotensive episodes between a proactive approach guided by the Hypotension Prediction Index (HPI) and a reactive goal-directed therapy (GDT) strategy guided by an advanced hemodynamic monitoring system. Results: The HPI group had a 65% reduction in hypotensive events (225 vs. 633, p < 0.001), with a 72% reduction in their duration (14 vs. 49 min, p < 0.001) and an 85% reduction in their severity (time-weighted average MAP < 65 mmHg 0.11 vs. 0.76, p < 0.001) compared to the GDT group. The HPI-guided group showed a reduction in postoperative infectious complications (10 vs. 26) and in-hospital length of stay (8 ± 4 versus 13 ± 8 days). Conclusions: A proactive approach may allow attenuating the occurrence and severity of hypotensive events more than a reactive goal-directed approach during radical cystectomy. Full article

Evaluation of socio-ecological vulnerability is crucial for sustainable management in mining cities. This study selected Xintai City, China, as a case and constructed a comprehensive vulnerability assessment framework based on 2010–2020 multi-source data. By integrating the Geodetector, spatial autocorrelation analysis, and ordered weighted averaging (OWA), we systematically explored the spatio-temporal patterns and driving mechanisms of socio-ecological vulnerability. The Theil index at the village level revealed finer spatial heterogeneity than large-scale analyses. The results show the following: (1) Socio-ecological vulnerability in Xintai City is generally moderate, with high-vulnerability areas concentrated in the urban center and former coal mining zones. Over the past decade, high—vulnerability levels in these areas have improved, whereas the urban-rural fringe has experienced a significant increase in vulnerability, primarily driven by industrial transfer and uneven resource allocation. (2) Geodetector analysis indicated a shift in dominant drivers from natural to socio-economic factors, with population density and construction land proportion surpassing natural conditions such as average annual rainfall by 2020. Additionally, mining land proportion, land use change, and the spatial distribution of social services played key roles in shaping vulnerability patterns, while ecological and public service factors showed weaker explanatory power. (3) Scenario simulation based on OWA demonstrated that an economic-priority pathway leads to the outward expansion of vulnerable areas into mountainous regions, while an ecological-priority approach promotes spatial contraction and optimization of vulnerability zones. These findings provide scientific guidance for identifying key vulnerable areas and formulating differentiated management strategies, offering reference value for the sustainable development of resource-exhausted mining cities. Full article

Falls from height are a critical safety concern in the construction industry, underscoring the need for effective identification of high-risk worker behaviors near hazardous edges for proactive accident prevention. This study aimed to address this challenge by developing an improved action recognition model. We propose a novel dynamic spatio-temporal graph convolutional network (CoG-STGCN) that incorporates a center of gravity (CoG)-aware mechanism. The method computes global and local CoG using anthropometric priors and extracts four key dynamic CoG features, which a Multi-Layer Perceptron (MLP) then uses to generate modulation weights that dynamically adjust the skeleton graph’s adjacency matrix, enhancing sensitivity to stability changes. On a self-constructed dataset of eight typical edge-related hazardous behaviors, CoG-STGCN achieved a Top-1 accuracy of 95.83% (baseline ST-GCN: 93.75%) and an average accuracy of 94.17% in fivefold cross-validation (baseline ST-GCN: 92.91%), with significant improvements in recognizing actions involving rapid CoG shifts. The CoG-STGCN provides a more effective and physically informed approach for intelligent unsafe behavior recognition and early warning in built environments. Full article

Background: The Induced Membrane Technique (IMT), commonly known as the Masquelet Technique (MT), has shown promising results in the reconstruction of bone defects caused by osteomyelitis. However, it is not a standardized surgical protocol but a treatment concept that has undergone various modifications, often yielding heterogeneous outcomes. Methods: This retrospective, single-center clinical cohort study included 49 patients treated with the Bone Block Technique (BBT) between 2013 and 2019 for bone defects resulting from osteomyelitis. The primary outcomes were time to bone healing, reinfection rate, and time to full weight-bearing. Additionally, infectious disease parameters, surgical site complications (SSCs), and epidemiological data were evaluated. Results: Data from 49 patients (mean age: 51 years, range: 17.6–76.9; 28.6% female) were analyzed, with a mean follow-up of 6.1 years (range: 4–10.5). The average bone defect length was 4.2 cm (range: 2.1–8.4 cm), predominantly involving the lower extremity. Primary bone consolidation was achieved in 93%, and secondary consolidation (requiring additional surgery) in 7%. Revision surgery due to recurrent infection was necessary in 16.6% of cases. The average time to full weight-bearing was 101.3 days. Conclusions: The BBT, as a modified approach based on the original IMT, represents a viable and reproducible option for bone defect reconstruction. When applied in accordance with the principles of the Diamond Concept, this technique facilitates reliable primary consolidation with a low complication rate. Full article

High-precision satellite clock bias (SCB) prediction is essential for GNSS applications, including real-time precise point positioning (RT-PPP), Earth observation, planetary exploration, and spaceborne geodetic missions. However, during communication outages or when real-time SCB products are unavailable, RT-PPP may fail due to missing clock corrections. This underscores the necessity of reliable short-term SCB prediction in data-denied environments. To address this challenge, a hybrid model that integrates wavelet transform, a particle swarm optimization-enhanced gray model, and a first-order weighted local method is proposed for short-term SCB prediction. First, the novel model employs the db1 wavelet to perform three-level multi-resolution decomposition and single-branch reconstruction on preprocessed SCB, yielding one trend term and three detailed terms. Second, the particle swarm optimization algorithm is adopted to globally optimize the parameters of the traditional gray model to avoid falling into local optima, and the optimization-enhanced gray model is applied to predict the trend term. For the three detailed terms, the embedding dimension and time delay are calculated, and they are constructed in phase space to establish a first-order weighted local model for prediction. Third, the final SCB prediction is obtained by summing the predicted results of the trend term and the three detailed terms correspondingly. The BDS-3 SCB products from the GNSS Analysis Center of Wuhan University (WHU) are selected for experiments. Results indicate that the proposed model surpasses conventional linear polynomial (LP), quadratic polynomial (QP), gray model (GM), and Legendre (Leg.) polynomial models. The average precision and stability improvements reach (80.00, 79.16, 82.14, and 72.22) % and (36.36, 41.67, 41.67, and 61.11) % for 30 min prediction, (79.31, 78.57, 80.65, and 76.92) % and (44.44, 44.44, 47.37, and 74.36) % for 60 min prediction, and the average precision of the predicted SCB products is better than 0.20 ns and 0.21 ns for 30 min and 60 min, respectively. Furthermore, the proposed model exhibits strong robustness and is less affected by changes in clock types and the amount of modeling data. Therefore, in practical applications, the short-term SCB products predicted by the novel model are fully capable of satisfying the requirements of centimeter-level RT-PPP for clock bias precision. Full article

Automatic Modulation Classification (AMC) is vital for adaptive wireless communication, yet it faces challenges in complex environments, including insufficient feature extraction, feature redundancy, and high interclass similarity among modulation schemes. To address these limitations, this paper proposes the Multiscale Complex Convolution Spatiotemporal Attention Network (MCCSAN). In this work, we propose three key innovations tailored for AMC tasks: a multiscale complex convolutional module that directly processes raw I/Q sequences, preserving critical phase and amplitude information while extracting diverse signal features. A spatiotemporal attention mechanism dynamically weights temporal steps and feature channels to suppress redundancy and enhance discriminative feature focus. A combined loss function integrating cross-entropy and center loss improves intraclass compactness and interclass separability. Evaluated on the RML2018.01A dataset and RML2016.10A across SNR levels from −6 dB to 12 dB, MCCSAN achieves a state-of-the-art classification accuracy of 97.03% (peak) and an average accuracy improvement of 3.98% over leading methods. The study confirms that integrating complex-domain processing with spatiotemporal attention significantly enhances AMC performance. Full article

Unsupervised feature selection aims to characterize the cluster structure of original features and select the optimal subset without label guidance. However, existing methods overlook fuzzy information in the data, failing to model cluster structures between data effectively, and rely on squared error for data reconstruction, exacerbating noise impact. Therefore, a robust unsupervised feature selection algorithm based on fuzzy anchor graphs (FWFGFS) is proposed. To address the inaccuracies in neighbor assignments, a fuzzy anchor graph learning mechanism is designed. This mechanism models the association between nodes and clusters using fuzzy membership distributions, effectively capturing potential fuzzy neighborhood relationships between nodes and avoiding rigid assignments to specific clusters. This soft cluster assignment mechanism improves clustering accuracy and the robustness of the graph structure while maintaining low computational costs. Additionally, to mitigate the interference of noise in the feature selection process, an adaptive fuzzy weighting mechanism is presented. This mechanism assigns different weights to features based on their contribution to the error, thereby reducing errors caused by redundant features and noise. Orthogonal tri-factorization is applied to the low-dimensional representation matrix. This guarantees that each center represents only one class of features, resulting in more independent cluster centers. Experimental results on 12 public datasets show that FWFGFS improves the average clustering accuracy by 5.68% to 13.79% compared with the state-of-the-art methods. Full article

Background: Preterm infants and their families face a challenging experience during their stay in the neonatal intensive care unit (NICU). Family-centered care emphasizes the importance of welcoming parents, involving them in their baby’s daily care, and supporting the development of parenting skills. NICU staff should support parents in understanding their baby’s needs and in strengthening the parent–infant bond. Although many tools outline what parents should learn, there is a limited structured framework to monitor their involvement in the infant’s care. Tracking parental participation in daily caregiving activities could support professionals in effectively guiding families, ensuring a smoother transition to discharge. Aims: The aim of this study was to evaluate the adherence to and effectiveness of a structured tool for parental involvement in the NICU. This tool serves several key purposes: to track the progression and timing of parents’ autonomy in caring for their baby, to support parents in building caregiving competencies before discharge, and to standardize the approach of NICU professionals in promoting both infant care and family engagement. Methods: A structured template form for documenting parental involvement (“together TO-CARE template”, TTCT) was integrated into the computerized chart adopted in the NICU of Modena. Nurses were asked to complete the TTCT at each shift. The template included the following assessment items: parental presence; type of contact with the baby (touch; voice; skin-to-skin); parental involvement in care activities (diaper changing; gavage feeding; bottle feeding; breast feeding); and level of autonomy in care (observer; supported by nurse; autonomous). We evaluated TTCT uploaded data for very low birth weight (VLBW) preterm infants admitted in the Modena NICU between 1 January 2023 and 31 December 2024. Staff compliance in filling out the TTCT was assessed. The timing at which parents achieved autonomy in different care tasks was also measured. Results: The TTCT was completed with an average of one entry per day, during the NICU stay. Parents reached full autonomy in diaper changing at a mean of 21.1 ± 15.3 days and in bottle feeding at a mean of 48.0 ± 22.4 days after admission. The mean length of hospitalization was 53 ± 38 days. Conclusions: The adoption of the TTCT in the NICU is feasible and should become a central component of care for preterm infants. It promotes family-centered care by addressing the needs of both the baby and the family. Encouraging early and progressive parental involvement enhances parenting skills, builds confidence, and may help reduce post-discharge complications and readmissions. Furthermore, the use of a standardized template aims to foster consistency among NICU staff, reduce disparities in care delivery, and strengthen the support provided to families of preterm infants. Full article