Voluntary motor effort (VME)-driven neuromuscular electrical stimulation (NMES)-robots have demonstrated superior rehabilitative efficacy in post-stroke wrist–hand motor restoration compared with purely mechanical systems. However, their neuromodulatory mechanisms remain unclear. This study aimed to investigate cortical responses elicited by an electromyography (EMG)-driven NMES-robot assisting wrist–hand movements in individuals with chronic stroke, and to evaluate robot-nervous system interactions, using EMG to capture VME. Cortical responses during wrist–hand extension were recorded in participants with chronic stroke (n = 18) and unimpaired controls (n = 20) using functional near-infrared spectroscopy (fNIRS). The EMG-driven NMES-robot delivered somatosensory stimulation with different combinations of mechanical stimulation (MS) to the targeted joints and NMES to the forearm muscles, with or without VME. Combined MS and NMES co-stimulation activated wider cortical areas than MS or NMES alone in both groups. Compared with the unimpaired group, the stroke group exhibited greater ipsilesional activations in response to NMES (p < 0.05), while MS induced comparable activations in the contralateral hemisphere to the stimulated limb in both groups (p < 0.05). When VME was required, the stroke group exhibited wider contralesional involvement with MS only (p < 0.05) but reduced contralesional involvement during sole NMES (p < 0.05), compared to those in the co-stimulation. Additionally, the MS and NMES co-stimulation resulted in similar functional connectivity patterns between the stroke and unimpaired groups. Combined mechano-electrical somatosensory co-stimulation with MS and NMES to the affected wrist-hand, when voluntary motor effort is generated from the residual muscles, could elicit extensive cortical activation approaching normal patterns. Chinese Clinical Trial Registry with the identifier ChiCTR2200057839.

Tracing the tissue and cell-type origins of extracellular vesicles (EVs) in blood is critical for liquid biopsy and precision medicine, yet existing deconvolution methods remain limited by the need for labor-intensive reference signatures and poor adaptability to distribution shifts between tissue/cell-type datasets and EV transcriptomes. We introduce DADA-EV (Domain-Adaptive Diffusion Autoencoder for EVs), a hybrid deep learning framework that combines an autoencoder backbone with a generative simulation module and adversarial domain adaptation. DADA-EV features three key innovations: (1) a reference-free design that eliminates reliance on predefined signatures; (2) cross-domain generalization by aligning feature distributions between source (tissue/cell-type) and target (EV) domain; and (3) reduced dependence on source data during target-domain training. Extensive evaluations on pseudo-EV data show that DADA-EV consistently outperforms existing approaches, yielding accurate fraction estimates across diverse tissues and gene sets. Validation using in vitro cell-line mixtures further confirms its reliability in resolving complex compositions, demonstrating high sensitivity in detecting low-abundance targets. Applied to real EV transcriptomes, it reveals tissue- and cell-type heterogeneity across patient groups. In summary, DADA-EV provides a robust, reference-free, and generalizable solution for EV origin tracing, with strong potential to advance diagnosis, prognosis, and treatment monitoring via liquid biopsy.

This paper addresses the path planning challenge of underactuated Autonomous Underwater Vehicles (AUVs) in environments with ocean current disturbances and obstacles. A strategy is proposed that combines Double Deep Q-Network (DDQN) with a Dual-Branch Convolutional Neural Network (CNN): one branch processes the sonar occupancy grid for obstacle perception, while the other encodes Acoustic Doppler Current Profiler (ADCP) current measurements. This dual-stream design enables multimodal local perception without global prior information. A distance-gated reward encourages active exploitation of favorable currents, and quintic polynomial smoothing ensures kinematically feasible heading transitions for the underactuated platform. Simulation and hardware experimental results demonstrate that the proposed strategy adapts to ocean current disturbances, completing path planning tasks with autonomous obstacle avoidance.

Computational musculoskeletal models possess great potential to quantify biomechanics across joint, muscle, and ligament levels during dynamic movements - a capability essential for uncovering the underlying mechanisms of anterior cruciate ligament (ACL) injury. However, comprehensive frameworks that fully integrate these multilevel features remain scarce. In this study, we developed a novel computational musculoskeletal model by integrating a discrete-element knee model containing ligaments into a full-body musculoskeletal model featuring detailed trunk and lower-extremity musculature. The accuracy of estimated ligament and muscle geometries were verified against those of the original component models under identical prescribed motions, and the mechanical behavior of the knee ligaments was rigorously validated through a series of forward dynamics simulations. The validated model was then applied to simulate four high-risk movements in fifteen healthy participants, after which linear regression analysis was performed to quantify the associations among kinematic variables derived from inverse kinematics, muscle forces estimated using a personalized EMG-driven modeling approach, and ACL strain/force. The proposed model demonstrated robust knee joint stability across various loading conditions and flexion angles, while preserving accurate ligament and muscle geometries relative to the source models. Increased ACL strain during the landing phase was significantly associated with greater knee abduction and anterior tibial translation. Quadriceps, gastrocnemius, and adductor forces consistently exhibited ACL-loading effects across all tasks, whereas hamstring forces demonstrated a task-dependent relationship with ACL force. Ultimately, this model provides a powerful tool for identifying risk factors associated with ACL injury, facilitating the development and refinement of evidence-based prevention strategies.

This paper studies the path-following problem of an underactuated autonomous underwater vehicle (AUV) with the ocean current disturbances. A line-of-sight (LOS) guidance law is employed at the outer loop, while a Lyapunov-based model predictive controller (LMPC) is developed at the inner loop to ensure that the AUV can accomplish the path-following task under ocean current disturbances. The proposed LOS-LMPC inherits the stability and robustness of the extended state observer (ESO)-based auxiliary control law and utilizes online optimization to enhance the path-following performance of the AUV system. Simulations and hardware experiments conducted on the “Qilin” AUV demonstrate the effectiveness of the proposed method.

For a 3D path-following problem of a novel autonomous underwater vehicle (AUV), traditional strategies fail to effectively consider error constraints and the actual thruster configuration and saturation. The objective of this paper is to address these issues. We propose a hierarchical strategy consisting of an outer-loop Error-Constrained Line-of-Sight (LOS) strategy and an inner-loop model predictive control (MPC) method. The main contributions of this strategy are: 1) it considers error constraints and can be adjusted to reduce to a conventional LOS through parameter tuning, and 2) it explicitly accounts for thruster configuration and saturation. Finally, simulations demonstrate that this strategy effectively reduces path-following errors compared to traditional LOS+MPC while satisfying thruster configuration and saturation constraints with nearly identical thrust. Full-scale experiments validate the proposed control strategy.

This paper investigates underwater target tracking problems for autonomous underwater glider (AUG) swarms with the special mode of gliding motion and relative slow speeds. An underwater target tracking method for AUG swarms based on model predictive control (MPC) strategy and adaptive large neighborhood search (ALNS) algorithm is proposed. Firstly, the underwater target tracking problem for AUG swarms is introduced and the corresponding mathematical model is also established consisting of AUG model, target model, and the objective function of the target tracking. Secondly, the proposed MPC-ALNS based underwater target tracking method is presented with MPC-based dynamic tracking framework and ALNS based optimization model for AUG swarms. The MPC framework is employed to cope with the dynamic factors e.g., target maneuver scenarios in the underwater target tracking process. Within the MPC framework, an ALNS algorithm is tailored for solving the underwater target tracking problem of AUG swarms by designing dynamic operators. Finally, simulation experiments show that the proposed underwater target tracking method achieves superior tracking performance while maintaining comparable overall computational efficiency compared with the three representative method.

In order to address the challenges of communication and computation limitations as well as constraint satisfaction in underwater manipulators, this paper proposes a dual-mode event-triggered Nonlinear Model Predictive Control (ENMPC) strategy. First, tightened state constraints are designed to ensure robust constraint satisfaction under bounded disturbances. A dual-mode event-triggered mechanism is then introduced, which explicitly accounts for the lower bound of inter-event times to accommodate the limitations of onboard communication and computation capabilities, while simultaneously reducing the computational load. Second, the recursive feasibility and closed-loop stability of the proposed strategy are rigorously analyzed. Numerical simulations and hardware experiments demonstrate that the proposed strategy can effectively accomplish the control task while satisfying control constraints, state constraints, communication requirements, and computational limitations.

This study presents a modular and scalable wearable functional near-infrared spectroscopy (fNIRS) system for high-resolution cerebral hemodynamic signal acquisition. The system is based on compact optoelectronic modules and supports mixed measurements using short-separation and long-separation channels, offering good scalability and spatial adaptability. The integrated quartz light guide structure improves optical coupling efficiency between the probe and scalp. A series of in vivo experiments validated system performance. In a forearm arterial occlusion experiment, the system accurately captured concentration changes in oxygenated and deoxygenated hemoglobin during blood flow blockade and reperfusion, with large effect sizes (Cohen’s d > 0.9). In a prefrontal cortex Valsalva experiment, the biphasic response characteristic of neurovascular coupling was successfully resolved. In a 2-back working memory task, the system identified a task-related frequency component (0.0227 Hz) and right-lateralized prefrontal cortex activation (p = 0.023). These results demonstrate that the system exhibits a good signal-to-noise ratio and temporal dynamic response, enabling high-resolution mapping of regional hemodynamic changes. This work provides an effective solution for the development of wearable, modular, and high-precision multi-channel fNIRS systems.

Modeling underwater manipulators is challenging due to complex hydrodynamic effects and environmental uncertainties. This paper presents a data-driven control framework that combines Koopman operator theory with implicit robust model predictive control (MPC) for a newly designed three-degree-of-freedom(3-DOF) underwater manipulator. The Koopman operator enables global linearization of nonlinear dynamics through state lifting, allowing the construction of control-oriented linear models directly from experimental data without requiring explicit physical parameter identification. To address modeling errors and external disturbances inherent in the Koopman approximation, we propose an implicit robust Koopman-based MPC (r-KMPC) strategy. Unlike conventional tube-based robust MPC methods that require explicit construction of robust positively invariant sets via computationally intractable Minkowski sum operations, the proposed approach employs support function evaluations to implicitly represent constraint tightening, thereby enabling robust control for high-dimensional lifted systems. Theoretical guarantees on recursive feasibility and input-to-state stability(ISS) are established. The effectiveness and robustness of the proposed r-KMPC are validated through both simulations and field experiments on a “Qilin” underwater manipulator at 12 m depth, demonstrating superior tracking accuracy and robust constraint satisfaction compared to nominal Koopman-based MPC(KMPC).

To address the challenges of underwater target tracking in dynamic and turbid environments, this paper proposes a Sequential State Estimation-based Tracking (SSET) algorithm that integrates temporal and spatial information for robust performance. The SSET algorithm comprises three key components: a Kalman filter-based sequential state estimation module to predict target motion and establish temporal correlations, a score head module to evaluate template reliability and optimize predictions, and a mixed-sequential-state transformer (MSST) to fuse triplet features for spatio-temporal correlation. Evaluations on open-air and underwater benchmarks demonstrate SSET’s superiority: it achieves 91.7% precision and 69.8% success rate on terrestrial datasets, outperforming state-of-the-art methods by 0.6% and 0.2%, respectively. In underwater scenarios, SSET attains 56.5% precision and 55.8% success rate, with improvements in occlusion and low-resolution conditions. Underwater grasping experiments further validate its practicality, achieving a high success rate in controlled environments.

In this paper, an Autonomous Underwater Vehicle (AUV) swarm confrontation search method based on the Multi-Agent Deep Deterministic Policy Gradient (NIADDPG) algorithm is proposed for the AUV swarm confrontation search problem. Firstly, the AUV swarm confrontation search model is constructed. Secondly, the multi-agent state space, action space, observation space, and reward function are designed for the AUV swarm confrontation search problem, and the MADDPG algorithm is adopted to obtain the AUV swarm confrontation search method. Finally, the effectiveness and the generality of the proposed AUV swarm confrontation search method are verified by simulation experiments for the AUV swarm confrontation search tasks under different speed conditions and sonar detection distances.

BACKGROUND:Blood flow restriction creates a state with increased motor function that permits treatment modalities to induce muscle hypertrophy. Blood flow-restricted exercise training (BFRET) may induce motor learning and boost the facilitatory effect of exercise training (ET). OBJECTIVE:This study investigated the effects of BFRET on post-stroke hemi paretic lower extremity function and walking capacity recovery. METHODOLOGY:This randomized clinical trial was conducted from September 2021 to October 2022 at the Department of Rehabilitation Medicine of the Second Affiliated Hospital of Chongqing Medical University in China. Participants were randomized 1:1 to BFRET or ET, each involving 30 minutes of training twice per day for 4 weeks. MAIN OUTCOMES MEASURES:The main outcomes were manual muscle testing (MMT) and Fugl-Meyer assessment scale-lower extremity (FMA-LE), the timed up and go test (TUGT), Outcomes were assessed by blinded raters after 4 weeks of training. RESULTS:40 participants mean [SD] age 48.79[12.58] years, 30 males [75%], 20 were randomized to BFRET and 20 to ET. The mean (SD) time since stroke was 2.5 (1.3) years. The MMT scores showed greater strength by within-group comparisons and superior changes in hip flexion and plantar flexion in the BFRET group. CONCLUSIONS:BFRET is superior to ET alone in enhancing muscle mass and strength in the lower extremities. BFRET may improve the function of the lower extremities through physiological adaptations for muscle hypertrophy. REGISTRATION:URL: https://www.clinicaltrials.gov; Unique identifier: ChiCTR2100050206.

This paper presents an innovative control scheme combining implicit rigid tube model predictive control (IRTMPC) with adaptive sliding mode control (ASMC) for the trajectory tracking of underwater manipulators, addressing the challenges posed by model uncertainties and external disturbances in underwater environment. The IRTMPC approach avoids the complexity of explicit set algebra operations by utilizing implicit minimal robust invariant set representations, enabling efficient computation suitable for high-dimensional systems. The auxiliary tube control law in IRTMPC is designed by discrete ASMC, reinforcing the system's stability and precision. Meanwhile, the dynamic parameters of a newly designed underwater manipulator are identified using a least squares identification method based on the integral operator. Finally, the effectiveness of the proposed IRTMPC-ASMC scheme for controlling the underwater manipulator is validated through trajectory tracking control simulation and experiments.

This paper proposes a convolutional multi-agent deep deterministic policy gradient method with prioritized experience replay (PER-CMADDPG) for the problem of multi-AUV cooperative search for moving targets. A comprehensive mathematical model of the multi-AUV cooperative search for moving targets is first established, which includes the environment model, the AUV model, and the information update and fusion model. Building upon the MADDPG framework, the proposed PER-CMADDPG method introduces two major enhancements. Convolutional neural networks (CNNs) are integrated into both the actor and critic networks to extract spatial features from local observation maps and global states, enabling agents to better perceive the spatial structure of the environment. In addition, a prioritized experience replay (PER) mechanism is incorporated to improve learning efficiency by emphasizing informative experiences during training, thereby accelerating policy convergence. Simulation experiments demonstrate that the proposed method achieves faster convergence and higher rewards compared with MADDPG. Furthermore, the influences of the multi-AUV cluster system’s scale, AUV speed, and sonar detection radius on performance are analyzed. The results verify the effectiveness of the proposed PER-CMADDPG method for the multi-AUV cooperative search for moving targets.

The study aimed to investigate changes in corticomuscular coupling during elbow flexion and extension in stroke survivors using functional near-infrared spectroscopy (fNIRS) and surface electromyography (sEMG), and to evaluate the relationship between coupling characteristics and clinical assessment scales. This study recruited 12 stroke survivors and 12 age-matched healthy subjects, and further divided the subjects into the affected side group, healthy-side group and age-matched healthy group. They performed elbow flexion and extension tasks at 30% and 70% of the maximum voluntary contraction (MVC). The cerebral blood flow dynamics of the bilateral prefrontal cortex, motor cortex, and occipital lobe, along with sEMG signals from the biceps brachii and triceps brachii, were simultaneously recorded. At matched force levels, the fuzzy approximate entropy values of both agonist and antagonistic muscles were notably lower in the affected group compared to the healthy group (P < 0.05). The effective connectivity from the ipsilateral motor cortex to the contralateral motor cortex during elbow movements in the affected group showed a meaningful positive association with the Fugl-Meyer Assessment (FMA) scale. Additionally, the transfer entropy from the contralateral motor cortex to the agonist muscle in the affected group demonstrated a significant positive correlation with the FMA scale at 70% MVC during elbow flexion. This research identified differences in intermuscular coordination, brain network connectivity, and corticomuscular coupling between stroke survivors and healthy individuals during motor tasks and our findings suggest that it can serve as a potential quantitative marker for assessing upper limb motor function post-stroke. The relationship between these characteristics and clinical scales signifies potential quantitative assessment parameters for stroke rehabilitation, underscoring the importance of exploring corticomuscular coupling in the recovery of upper limb motor function post-stroke.

Accurate delineation of seabed sedimentary structures is vital for marine engineering and exploration. Sub-bottom profilers (SBPs) provide a cost-effective means of sub-surface imaging; however, their utility is often hampered by noise, multiple reflections, and data inconsistencies. This study introduces an automated, high-precision algorithm for extracting layer boundaries from SBP images. The algorithm employs a three-step computational framework: (1) adaptive preprocessing involving SEG-Y decoding, image reconstruction, and hybrid noise reduction using the Maximum Inter-class Distance method; (2) gradient-based edge detection utilizing localized statistical windows and dynamic thresholding techniques; and (3) topology-aware refinement through connectivity tracking, neighborhood merging, and false boundary elimination. Sea trial results demonstrate the robustness of the proposed method, achieving enhanced continuity and precision in automated interpretation, thus providing a reliable tool for marine geological structure analysis.

This paper presents a fault-tolerant model predictive control approach for cross-rudder autonomous underwater vehicles to achieve heading control, considering rudder stuck faults and unknown disturbances. Specifically, additive faults in the rudders are addressed, and an active fault-tolerant control strategy is employed. Fault models of autonomous underwater vehicles have been established to develop the fault-tolerant control method. In the controller design, the stuck faults of complete rudder failure are incorporated to ensure the heading angle control of the autonomous underwater vehicle in faulty conditions. Furthermore, the fault term is decoupled from the control input, and the decoupled control input, along with corresponding constraints, is incorporated into the model’s predictive controller design. This approach facilitates controller reconfiguration, thereby enhancing and optimizing control performance. Simulation results demonstrate that the proposed fault-tolerant model predictive control method can effectively achieve stable navigation and heading adjustment under rudder fault conditions in autonomous underwater vehicles.

Extracellular vesicles (EVs), considered as a form of liquid biopsy, have gained significant attention in recent years due to their stability and the preservation of disease markers. Research studies underscore the clinical significance of molecules found in EVs, highlighting their role as communicative mediators between cells. However, analyzing this data is challenging due to noisy measurements, having far more variables than samples, and some groups (e.g., disease subtypes or experimental conditions) having much less data than others. We therefore develop an algorithm to address aforementioned challenges for the classification of imbalanced EVs omics data. We propose the EV Meta-Weight Elastic Net Algorithm (MWENA), which utilizes logistic regression with elastic net regularization for the classification and identification of EV signatures, effectively addressing the challenges posed by high-dimensional small sample sizes. To mitigate issues related to class imbalance and high noise levels, MWENA incorporates an automatic sample re-weighting function, which uses a meta-net to adaptively learn generalizable patterns directly from the data itself. We validate the MWENA algorithm on both simulated data and EVs omics data, covering six classification tasks that involve four different types of diseases (pancreatic ductal adenocarcinoma, interstitial lung diseases, colorectal cancer, and ovarian cancer) and three clinical scenarios (disease diagnosis, disease-stage screening, and disease-subtype classification). Compared to other machine learning methods, MWENA demonstrates superiority in identifying small class samples and achieves the highest scores in both sensitivity and G-means. Biological analysis is also performed to further explore the significance of selected signatures as biological markers and their roles in disease mechanisms. We anticipate that our proposed approach will take a modest step in harnessing EV omics data to discover biomarkers, aiding researchers in gaining a comprehensive understanding of biological processes.

The integration of digital pathology images and genetic data is a developing field in cancer research, presenting potential opportunities for predicting survival and classifying grades through multiple source data. However, obtaining comprehensive annotations proves challenging in practical medical settings, and the extraction of features from high -resolution pathology images is hindered by inter-domain disparities. Current data fusion methods ignore the spatio-temporal incongruity among multimodal data. To address the above challenges, we propose a novel self-supervised transformer-based pathology feature extraction strategy, and construct an interpretable Progressive Multimodal Fusion Network (PMFN-SSL) for cancer diagnosis and prognosis. Our contributions are mainly divided into three aspects. Firstly, we propose a joint patch sampling strategy based on the information entropy and HSV components of an image, which reduces the demand for sample annotations and avoid image quality degradation caused by manual contamination. Secondly, a self-supervised transformerbased feature extraction module for pathology images is proposed and innovatively leverages partially weakly supervised labeling to align the extracted features with downstream medical tasks. Further, we improve the existing multimodal feature fusion model with an progressive fusion strategy to reduce the inconsistency between multimodal data due to differences in collection of temporal and spatial. Abundant ablation and comparison experiments demonstrate that the proposed data preprocessing method and multimodal fusion paradigm strengthen the quality of feature extraction and improve the prediction based on real cancer grading and prognosis. Code and trained models are made available at: https://github.com/Mercuriiio/PMFN-SSL.