Search Results (13,930)

The reliability of molecular diagnostic and prognostic tools is contingent on the quality of biospecimens, which are often collected during surgical procedures. This study investigated the impact of surgical manipulation on gene expression in the urinary bladder mucosa during radical cystectomy. Seventeen patients with urinary bladder cancer were enrolled, and paired pre- and post-surgery biopsies were analyzed. Pre-surgical biopsies were obtained in situ under anesthesia, while post-surgical biopsies were collected ex vivo following bladder removal. Total RNA was extracted, and gene expression was assessed using qPCR arrays, measuring the expression of 374 inflammation-related genes. The findings from the exploratory phase were further validated by analyzing key genes in an independent patient cohort using TaqMan^® gene-specific assays. Exploratory analysis revealed significant differential expression in 27 genes, with key genes such as IL6, FOS, and PTGS2 being upregulated post-surgery. Validation of five selected genes in an independent cohort confirmed these findings. This study reinforces the necessity of accounting for surgery-induced alterations in gene expression when analyzing tissue samples collected intraoperatively. By elucidating the molecular impact of surgical interventions, this work provides critical insights for refining experimental methodologies and enhancing the interpretability of gene expression studies in clinical and research settings. Full article

Accurate and timely detection of postural control impairments in stroke patients is crucial for effective rehabilitation and fall prevention. Traditional clinical assessments often rely on qualitative observation and handcrafted features, which may fail to capture the nonlinear and uncertain nature of postural deficits. This study addresses these limitations by introducing a hybrid deep learning framework that integrates Convolutional Neural Networks (CNNs) with Type-2 fuzzy logic activation to robustly classify sensory dysfunction under altered balance conditions. Using an EquiTest-derived dataset of 8316 labeled samples from 700 participants across six standardized sensory manipulation scenarios, the proposed method achieved 97% accuracy, 96% precision, 97% sensitivity, and 96% specificity, outperforming conventional CNN and other baseline classifiers. The approach demonstrated resilience to measurement noise down to 1 dB SNR, confirming its robustness in realistic clinical environments. These results suggest that the proposed system can serve as a practical, non-invasive tool for clinical diagnosis and personalized rehabilitation planning, supporting data-driven decision-making in stroke care. Full article

With the increasing deployment of large language models (LLMs) in diverse applications, security vulnerability attacks pose significant risks, such as prompt injection. Despite growing awareness, structured, hands-on educational platforms for systematically studying these threats are lacking. In this study, we present an interactive training framework designed to teach, assess, and mitigate prompt injection attacks through a structured, challenge-based approach. The platform provides progressively complex scenarios that allow users to exploit and analyze LLM vulnerabilities using both rule-based adversarial testing and Open Worldwide Application Security Project-inspired methodologies, specifically focusing on the LLM01:2025 prompt injection risk. By integrating attack simulations and guided defensive mechanisms, this platform equips security professionals, artificial intelligence researchers, and educators to understand, detect, and prevent adversarial prompt manipulations. The platform highlights the effectiveness of experiential learning in AI security, emphasizing the need for robust defenses against evolving LLM threats. Full article

Deepfake audio refers to speech that has been synthetically generated or altered through advanced neural network techniques, often with a degree of realism sufficient to convincingly imitate genuine human voices. As these manipulations become increasingly indistinguishable from authentic recordings, they present significant threats to security, undermine media integrity, and challenge the reliability of digital authentication systems. In this study, a robust detection framework is proposed, which leverages the power of self-supervised learning (SSL) and attention-based modeling to identify deepfake audio samples. Specifically, audio features are extracted from input speech using two powerful pretrained SSL models: HuBERT-Large and WavLM-Large. These distinctive features are then integrated through an Attentional Multi-Feature Fusion (AMFF) mechanism. The fused features are subsequently classified using a NeXt-Time Delay Neural Network (NeXt-TDNN) model enhanced with Efficient Channel Attention (ECA), enabling improved temporal and channel-wise feature discrimination. Experimental results show that the proposed method achieves a 0.42% EER and 0.01 min-tDCF on ASVspoof 2019 LA, a 1.01% EER on ASVspoof 2019 PA, and a pooled 6.56% EER on the cross-channel ASVspoof 2021 LA evaluation, thus highlighting its effectiveness for real-world deepfake detection scenarios. Furthermore, on the ASVspoof 5 dataset, the method achieved a 7.23% EER, outperforming strong baselines and demonstrating strong generalization ability. Moreover, the macro-averaged F1-score of 96.01% and balanced accuracy of 99.06% were obtained on the ASVspoof 2019 LA dataset, while the proposed method achieved a macro-averaged F1-score of 98.70% and balanced accuracy of 98.90% on the ASVspoof 2019 PA dataset. On the highly challenging ASVspoof 5 dataset, which includes crowdsourced, non-studio-quality audio, and novel adversarial attacks, the proposed method achieves macro-averaged metrics exceeding 92%, with a precision of 92.07%, a recall of 92.63%, an F1-measure of 92.35%, and a balanced accuracy of 92.63%. Full article

Most amino acids contain a chiral center and thus, can exist as L- and D-isomers. For many years, it was thought that only the L-isomers were present in mammals. However, in recent decades it has been demonstrated that D-isomers are also present. Three of these amino acids, namely D-serine, D-aspartate, and D-alanine, have been proposed to play a role in the etiology of schizophrenia via interactions with glutamate receptors. D-Serine and D-alanine act at the glycine modulatory site on the NMDA receptor, while D-aspartate acts at the glutamate site on the same receptor. D-aspartate also acts on the mGlu5 receptor and can stimulate glutamate release presynaptically. Preclinical studies have reported that manipulations to reduce brain levels of D-serine, D-aspartate, or D-alanine lead to schizophrenia-relevant behaviors, and clinical studies have reported reduced levels of these D-amino acids in the brain tissue (postmortem) and/or body fluids from schizophrenia patients compared to those noted in controls, although there are some contradictory findings. The possible use of these amino acids and/or the manipulation of their relevant enzymes in the treatment of schizophrenia are described. D-Cysteine has been identified recently in human brain tissue, with the highest values in white matter; demonstration of its involvement in brain development has led to speculation that it could be involved in the etiology of schizophrenia, identifying it as a potential therapy in combination with antipsychotics. Future directions and potential problems that should be considered in studies on D-amino acids and their relevant enzymes in schizophrenia are discussed. Full article

Cauliflower, a widely cultivated vegetable crop valued for its edible curds, faces a persistent threat from insect pests, which are typically managed using synthetic insecticides. This study evaluated the benefits of intercropping practices as part of an ecological pest management strategy in cauliflower cultivation during the winter seasons of 2017–18 and 2021–22. Nine insect pests belonging to six families of three orders were recorded. The calendula intercropping system (IS) consistently showed the lowest infestation by Plutella xylostella and Pieris brassicae/plant. Calendula IS had attracted the highest numbers of syrphids, Cotesia glomerata, Diaeretiella rapae, Cotesia vestalis, and coccinellids such as Coccinella septempunctata and Cheilomenes sexmaculata. In candytuft IS, a strong tri-trophic interaction between the flower and D. rapae significantly reduced aphid populations, for each additional D. rapae, aphid numbers decreased by 48.53 in 2018. The marigold IS recorded the highest Shannon diversity index in 2021–22. The longest adult survival of C. septempunctata (8.67 ± 3.35 days), in the absence of aphids was recorded on candytuft flowers. The total sugars and protein in flowers positively influenced the longevity of the adult coccinellid beetles (R²-40.42 and 20.79%, respectively). Calendula intercropping yielded the highest revenue return of Indian rupee (₹) 11.33 per INR 1 invested, compared to the cauliflower monocrop (1.58). These findings demonstrate that, intercropping and habitat manipulation can enhance ecological pest control and reduce the dependence on synthetic chemicals. Full article

Graph Neural Networks (GNNs) have shown strong performance in link prediction, a core task in graph analysis. However, recent studies reveal their vulnerability to backdoor attacks, which can manipulate predictions stealthily and pose significant yet underexplored security risks. The existing backdoor strategies for link prediction suffer from two key limitations: gradient-based optimization is computationally intensive and scales poorly to large graphs, while single-node triggers introduce noticeable structural anomalies and local feature inconsistencies, making them both detectable and less effective. To address these limitations, we propose a novel backdoor attack framework grounded in the principle of homophily, designed to balance effectiveness and stealth. For each selected target link to be poisoned, we inject a unique path-based trigger by adding a bridge node that acts as a shared neighbor. The bridge node’s features are generated through a context-aware probabilistic sampling mechanism over the joint neighborhood of the target link, ensuring high consistency with the local graph context. Furthermore, we introduce a confidence-based trigger injection strategy that selects non-existent links with the lowest predicted existence probabilities as targets, ensuring a highly effective attack from a small poisoning budget. Extensive experiments on five benchmark datasets—Cora, Citeseer, Pubmed, CS, and the large-scale Physics graph—demonstrate that our method achieves superior performance in terms of Attack Success Rate (ASR) while maintaining a low Benign Performance Drop (BPD). These results highlight a novel and practical threat to GNN-based link prediction, offering valuable insights for designing more robust graph learning systems. Full article

Deformable object manipulation (DOM) is a primary bottleneck for the real-world application of autonomous robots, requiring advanced frameworks for sensing, perception, modeling, planning, and control. When fragile objects such as soft tissues or fruits are involved, ensuring safety becomes the paramount concern, fundamentally altering the manipulation problem from one of pure trajectory optimization to one of constrained optimization and real-time adaptive control. Existing DOM methodologies, however, often fall short of addressing fragility constraints as a core design feature, leading to significant gaps in real-time adaptiveness and generalization. This review systematically examines individual components in DOM with a focus on their effectiveness in handling fragile objects. We identified key limitations in current approaches and, based on this analysis, discussed a promising framework that utilizes both low-latency reflexive mechanisms and global optimization to dynamically adapt to specific object instances. Full article

The development of new types of nanocomposites capable of manipulating light is critical for various modern photonics applications. Recently, we proposed the use of overlooked glass-forming ionic liquid crystals made of cadmium octanoate containing gold, carbon, or both carbon and gold nanoparticles as promising optical and nonlinear optical materials. These were characterized using nanosecond laser pulses at a wavelength of 532 nm. In this paper, femtosecond radiation at different wavelengths (600 nm and 800 nm) is employed to study ultrafast electronic nonlinear optical processes in mesomorphic glass nanocomposites. The observed nonlinear optical response probed at the femtosecond time scale dramatically differs from that at the nanosecond time scale reported previously. The intensity-dependent effective nonlinear absorption coefficient of all studied samples remains positive due to the dominant reverse saturable absorption effect, while the nonlinear refractive index exhibits a sign reversal depending on the intensity and wavelength of laser pulses. The strategy for producing glass-forming ionic liquid crystal gold–carbon nanocomposites with an ultrafast nonlinear optical response is of high interest for modern applications in advanced photonics, and it can also be applied to other types of glass-forming metal alkanoates and nanomaterials. Full article

This study presents a novel, fully passive radiofrequency (RF)-based localization system designed to detect the position of a robotic hand on a flat surface within its tactile range, particularly in scenarios where other sensing systems may face limitations. The system employs U-shaped, chipless resonator tags printed on the surface using a customized conductive ink, together with a coplanar RF probe integrated into the robotic hand, to determine position through impedance variations. Unlike conventional approaches, the proposed method provides a compact, low-cost, and robust solution that is resilient to variations in lighting, dust, and other environmental conditions. The resonator tags are arranged in a structured grid inspired by a Sudoku pattern, enabling both position and orientation detection in the near-field region. The system is fabricated on 3D-printed flexible substrates using a flexible and stretchable conductive ink, and its performance is validated through both electromagnetic simulations and experimental measurements. The results confirm that the proposed approach enables accurate and repeatable two-dimensional localization of the robotic hand under various configurations. This work introduces a scalable, high-precision, and vision-independent sensing platform with strong potential for robotic manipulation in challenging environments. Full article

Background: Cognitive biases in information processing, particularly attentional and memory biases, play a crucial role in the development and maintenance of Major Depressive Disorder (MDD). These biases lead individuals with MDD to preferentially attend to and remember negative information, thereby maintaining a depressed mood. A recently proposed attentional resources model suggests that exposure to negative stimuli leads to deeper cognitive processing of subsequent information, regardless of its content. Based on this model, the current study investigated a novel paradigm that manipulated exposure to negative emotional stimuli and examined its effect on information processing and mood improvement. Method: Thirty-eight unmedicated participants with MDD and no comorbid disorders, and 37 healthy controls, completed three blocks of an emotional recall task, which involved watching a short emotional video followed by a recall task of neutral or positive valence stories. Mood changes were assessed throughout the task. Results: Results revealed that both the MDD and HC groups reported improved mood after exposure to a negative emotional video followed by a positive story. Conclusions: These results have important clinical implications. The paradigm may be applied in a broader sense as an active tool that may help to improve mood in depression treatment. Full article

Agriculture is facing increasing challenges due to labor shortages, rising productivity demands, and the need to operate in unstructured environments. Robotics, particularly soft robotics, offers promising solutions for automating delicate tasks such as fruit harvesting. While numerous soft grippers have been proposed, most focus on grasping and lack the capability to detach fruits with rigid peduncles, which require cutting. This paper presents a novel modular hexagonal soft gripper that integrates soft pneumatic actuators, embedded mechano-optical force sensors for real-time contact monitoring, and a self-centering iris-type cutting mechanism. The entire system is 3D-printed, enabling low-cost fabrication and rapid customization. Experimental validation demonstrates successful harvesting of bell peppers and identifies cutting limitations in tougher crops such as aubergine, primarily due to material constraints in the actuation system. This dual-capability design contributes to the development of multifunctional robotic harvesters capable of adapting to a wide range of fruit types with minimal requirements for perception and mechanical reconfiguration. Full article

This paper proposes optimal sliding mode fault-tolerant control for multiple robotic manipulators in the presence of external disturbances and actuator faults. First, a quantitative prescribed performance control (QPPC) strategy is constructed, which relaxes the constraints on initial conditions while strictly restricting the trajectory within a preset range. Second, based on QPPC, adaptive gain integral terminal sliding mode control (AGITSMC) is designed to enhance the anti-interference capability of robotic manipulators in complex environments. Third, a critic-only neural network optimal dynamic programming (CNNODP) strategy is proposed to learn the optimal value function and control policy. This strategy fits nonlinearities solely through critic networks and uses residuals and historical samples from reinforcement learning to drive neural network updates, achieving optimal control with lower computational costs. Finally, the boundedness and stability of the system are proven via the Lyapunov stability theorem. Compared with existing sliding mode control methods, the proposed method reduces the maximum position error by up to 25% and the peak control torque by up to 16.5%, effectively improving the dynamic response accuracy and energy efficiency of the system. Full article

We consider a multi-species mixture of interacting bosons, $N_1$ bosons of mass $m_1$, $N_2$ bosons of mass $m_2$, and $N_3$ bosons of mass $m_3$, in a harmonic trap with frequency $\omega$. The corresponding intra-species interaction strengths are ${\lambda }_{11}$, ${\lambda }_{22}$, and ${\lambda }_{33}$, and the inter-species interaction strengths are ${\lambda }_{12}$, ${\lambda }_{13}$, and ${\lambda }_{23}$. When the shape of all interactions is harmonic, the system corresponds to the generic multi-species harmonic-interaction model, which is exactly solvable. We start by solving the many-particle Hamiltonian and concisely discussing the ground-state wavefunction and energy in explicit forms as functions of all parameters, the masses, numbers of particles, and the intra-species and inter-species interaction strengths. We then explicitly compute the reduced one-particle density matrices for all the species and diagonalize them, thus generalizing the treatment by the authors earlier. The respective eigenvalues determine the degree of fragmentation of each species. As an application, we focus on phenomena that do not arise in the corresponding single-species or two-species systems. For instance, we consider a mixture of two kinds of bosons in a bath made by a third kind, controlling the fragmentation of the former by coupling to the latter. Another example exploits the possibility of different connectivities (i.e., which species interacts with which species) in the mixture, and demonstrates how the fragmentation of species 3 can be manipulated by the interaction between species 1 and species 2, when species 3 and 1 do not interact with each other. We highlight the properties of fragmentation that only appear in the multi-species mixture. Further applications are briefly discussed. Full article

Robotic manipulators in warehousing and logistics often face complex tasks that involve multiple steps, frequent task switching, and long-term dependencies. Inspired by the hierarchical structure of human motor control, this paper proposes a Hierarchical Reinforcement Learning (HRL) framework utilizing a multi-task goal-conditioned Decision Transformer (MTGC-DT). The high-level policy treats the Markov decision process as a sequence modeling task, allowing the agent to manage temporal dependencies. The low-level policy is made up of parameterized action primitives that handle physical execution. This design improves long-term reasoning and generalization. This method is evaluated on two common logistics manipulation tasks: sequential stacking and spatial sorting with sparse reward and low-quality dataset. The main contributions include introducing a HRL framework that integrates Decision Transformer (DT) with task and goal embeddings, along with a path-efficiency loss (PEL) correction and designing a parameterized, learnable primitive skill library for low-level control to enhance generalization and reusability. Experimental results demonstrate that the proposed Decision Transformer-based Hierarchical Reinforcement Learning (DT-HRL) achieves over a 10% higher success rate and over 8% average reward compared with the baseline, and a normalized score increase of over 2% in the ablation experiments. Full article