This study introduces a novel approach to robot-assisted ankle rehabilitation by proposing a dual-agent multiple model reinforcement learning (DAMMRL) framework, leveraging multiple model adaptive control (MMAC) and co-adaptive control strategies. In robot-assisted rehabilitation, one of the key challenges is modeling human behavior due to the complexity of human cognition and physiological systems. Traditional single-model approaches often fail to capture the dynamics of human-machine interactions. Our research employs a multiple model strategy, using simple submodels to approximate complex human responses during rehabilitation tasks, tailored to varying levels of patient incapacity. The proposed system's versatility is demonstrated in real experiments and simulated environments. Feasibility and potential were evaluated with 13 healthy subjects and nine patients with lower-limb motor disorders, yielding promising results that affirm the anticipated benefits of the approach. This study not only introduces a new paradigm for robot-assisted ankle rehabilitation but also opens the way for future research in adaptive, patient-centered therapeutic interventions.

Ultra-high molecular weight polyethylene (UHMWPE) has been widely used in total joint arthroplasty for orthopedic and spinal implants. However, the biological response to UHMWPE wear particles has been identified as a major contributor to inflammatory synovitis and periprosthetic osteolysis, which could lead to aseptic loosening and long-term implant failure. Traditional manual detection and classification of UHMWPE wear particles are labor-intensive, time-consuming, and prone to human error, which requires the development of automated detection techniques.This study proposes a novel deep learning-based framework for detecting UHMWPE wear particles, utilizing high-resolution field emission gun-scanning electron microscopy (FEG-SEM) images. The proposed approach employs an enhanced YOLOv9 object detection model, incorporating programmable gradient information (PGI) and generalized efficient layer aggregation networks (GELAN) to improve the localization and detection accuracy of small objects. Additionally, a customized Focal Loss function is integrated to address class imbalance and enhance sensitivity to submicron and nanoscale wear particles.Experimental evaluations demonstrate that our proposed model achieves a mean average precision (mAP) of 84.0%, outperforming the baseline YOLOv5 model by 7.7%. Furthermore, compared to mainstream object detection models such as YOLOv8 and Faster R-CNN, our approach exhibits superior detection accuracy and robustness, particularly in identifying wear particles in complex backgrounds and overlapping regions.In addition to developing an advanced detection algorithm, this study establishes a dedicated and expert-annotated UHMWPE wear particle dataset, addressing a critical gap in orthopedic implant research. The proposed framework provides a scalable, high-precision, and cost-effective solution for the automated detection of UHMWPE wear particles, supporting improved implant monitoring, osteolysis prevention, and clinical decision-making in orthopedic and spinal implant evaluations.

Distributed control systems offer significant benefits such as enhanced flexibility, improved failure tolerance, and simplified system design compared to centralized systems. While the controllability of linear square systems under decentralized structures incorporating integral action is well-explored, challenges arise when dealing with non-square processes, which are prevalent in various control and optimization scenarios. Traditionally, non-square systems are manipulated into square configurations by adjusting inputs and outputs to apply decentralized integral controllability (DIC) analysis, potentially affecting the system’s reliability and controller flexibility. To address this issue, we propose a direct application of decentralized integral controllability to non-square processes, termed DIC-NSQ. We approach this problem by representing the system through a state space description, transforming it into standard singular perturbation form, and subsequently establishing its necessary as well as sufficient conditions using singular perturbation analysis. Through numerical examples, we demonstrate that DIC-NSQ effectively maintains offset-free tracking and ensures robust performance, even in the presence of actuator failures. This contribution marks an advancement in the field of distributed control systems, particularly for applications involving non-square processes.

This paper focuses on the task of survival time analysis for lung cancer. Although much progress has been made in this problem in recent years, the performance of existing methods is still far from satisfactory. Traditional and some deep learning-based survival time analyses for lung cancer are mostly based on textual clinical information such as staging, age, histology, etc. Unlike existing methods that predicting on the single modality, we observe that a human clinician usually takes multimodal data such as text clinical data and visual scans to estimate survival time. Motivated by this, in this work, we contribute a smart cross-modality network for survival analysis network named Lite-ProSENet that simulates a human's manner of decision making. Extensive experiments were conducted using data from 422 NSCLC patients from The Cancer Imaging Archive (TCIA). The results show that our Lite-ProSENet outperforms favorably again all comparison methods and achieves the new state of the art with the 89.3% on concordance. The code will be made publicly available.

Brain-Computer Interfaces (BCIs) connect the brain to external control devices, necessitating the accurate translation of brain signals such as from electroencephalography (EEG) into executable commands. EEG MI classification has numerous applications, including neurorehabilitation for stroke patients, control of assistive robotic devices, and advancements in neurofeedback systems. Graph Neural Networks (GCN) have been increasingly applied for classifying EEG Motor Imagery (MI) signals, primarily because they incorporates the spatial relationships among EEG channels, resulting in improved accuracy over traditional convolutional methods. However, existing methods for constructing adjacency matrices, such as Geodesic distances, Pearson Correlation Coefficient (PCC), and others, often rely on predefined inter-channel relationships. These methods not only demand high computational resources during inference but often achieve limited performance accuracy, particularly for single time-point EEG MI classification where rapid interpretation is crucial. To address this, our paper introduces the EEG Graph Lottery Ticket (EEG_GLT) algorithm, an innovative technique for constructing adjacency matrices for EEG channels. This method does not require pre-existing knowledge of inter-channel relationships, and it can be tailored to suit both individual subjects and GCN model architectures. We conducted an empirical study with 20 subjects and six different GCN architectures to compare the performance of our EEG_GLT adjacency matrix against both Geodesic and PCC adjacency matrices on time-resolved EEG MI dataset, PhysioNet dataset. Our EEG_GLT method consistently exceeded performance accuracy benchmarks. Additionally, we compared our model with state-of-the-art models, achieving superior results. EEG_GLT algorithm marks a breakthrough in development of optimal adjacency matrices, effectively boosting both computational accuracy and efficiency, making it well-suited for single time point classification of EEG MI signals that demand intensive computational resources.

Early identification of internal carotid artery (ICA) system diseases is critical for preventing stroke and other cerebrovascular events. Traditional diagnostic methods rely heavily on clinician expertise and costly imaging, limiting accessibility. This study aims to develop an interpretable machine learning (ML) model using common carotid artery (CCA) features to predict ICA disease risk, enabling efficient screening. Clinical data from 1612 patients (806 high-risk vs. 806 low-risk ICA disease) were analyzed. CCA features-blood flow, intima-media thickness, internal diameter, age, and gender-were used to train five ML models. Model performance was evaluated via accuracy, sensitivity, specificity, AUC-ROC, and F1 score. SHAP analysis identified key predictors. The support vector machine (SVM) achieved optimal performance (accuracy, 84.9%; AUC, 92.6%), outperforming neural networks (accuracy, 81.4%; AUC, 89.8%). SHAP analysis revealed CCA blood flow (negative correlation) and intima-media thickness (positive correlation) as dominant predictors. This study demonstrates that CCA hemodynamic and structural features, combined with interpretable ML models, can effectively predict ICA disease risk. The SVM-based framework offers a cost-effective screening tool for early intervention, particularly in resource-limited settings. Future work will validate these findings in multi-center cohorts.

The key to robot-assisted rehabilitation lies in the design of the human-machine interface, which must accommodate the needs of both patients and machines. Current interface designs primarily focus on machine control algorithms, often requiring patients to spend considerable time adapting. In this paper, we introduce a novel approach based on the Cooperative Adaptive Markov Decision Process (CAMDPs) model to address the fundamental aspects of the interactive learning process, offering theoretical insights and practical guidance. We establish sufficient conditions for the convergence of CAMDPs and ensure the uniqueness of Nash equilibrium points. Leveraging these conditions, we guarantee the system's convergence to a unique Nash equilibrium point. Furthermore, we explore scenarios with multiple Nash equilibrium points, devising strategies to adjust both Value Evaluation and Policy Improvement algorithms to enhance the likelihood of converging to the global minimal Nash equilibrium point. Through numerical experiments, we illustrate the effectiveness of the proposed conditions and algorithms, demonstrating their applicability and robustness in practical settings. The proposed conditions for convergence and the identification of a unique optimal Nash equilibrium contribute to the development of more effective adaptive systems for human users in robot-assisted rehabilitation.

The calibration of MEMS triaxial gyroscopes is crucial for achieving precise attitude estimation for various wearable health monitoring applications. However, gyroscope calibration poses greater challenges compared to accelerometers and magnetometers. This paper introduces an efficient method for calibrating MEMS triaxial gyroscopes via only a servo motor, making it well-suited for field environments. The core strategy of the method involves utilizing the fact that the dot product of the measured gravity and the rotational speed in a fixed frame remains constant. To eliminate the influence of rotating centrifugal force on the accelerometer, the accelerometer data is measured while stationary. The proposed calibration experiment scheme, which allows gyroscopic measurements when operating each axis at a specific rotation speed, making it easier to evaluate the linearity across a related speed range constituted by a series of rotation speeds. Moreover, solely the classical least squares algorithm proves adequate for estimating the scale factor, notably streamlining the analysis of the calibration process. Extensive numerical simulations were conducted to analyze the proposed method's performance in calibrating a triaxial gyroscope model. Experimental validation was also carried out using a commercially available MEMS inertial measurement unit (LSM9DS1 from Arduino nano 33 BLE SENSE) and a servo motor capable of controlling precise speed. The experimental results effectively demonstrate the efficacy of the proposed calibration approach.

The successive loop closure control approach is popular in engineering practice due to its ease of use. However, in most cases, the successive loop closure approach, for each loop closure, is in the configuration of single-input and single-output, which limits the usage of the control of multi-variable processes with coupled effects. In this paper, we extended the traditional successive loop closure control approach to multi-variable processes. To further improve the practicalness of the proposed method, in each closure, the decentralized integral controller will be designed. A sufficient condition for the existence of a controller to ensure the decentralized unconditional stability of each loop is proposed, under which the tuning of the loop gains becomes easier. Illustrative example is proposed to demonstrate the effectiveness of the proposed approach.

With the rise in global meat consumption and chicken becoming a principal source of white meat, methods for efficiently and accurately determining the freshness of chicken are of increasing importance, since traditional detection methods fail to satisfy modern production needs. A non-destructive method based on machine vision and machine learning technology was proposed for detecting chicken breast freshness. A self-designed machine vision system was first used to collect images of chicken breast samples stored at 4 degrees C for 1-7 d. The Region of Interest (ROI) for each image was then extracted and a total of 700 ROI images were obtained. Six color features were extracted from two different color spaces RGB (red, green, blue) and HSI (hue, saturation, intensity). Six main Gray Level Co-occurrence Matrix (GLCM) texture feature parameters were also calculated from four directions. Principal Component Analysis (PCA) was used to reduce the dimension of these 30 extracted feature parameters for multiple features image fusion. Four principal components were taken as input and chicken breast freshness level as output. A 10-fold cross-validation was used to partition the dataset. Four machine learning methods, Particle and Naive Bayes Classifier (NBC), were used to establish a chicken breast freshness level prediction model. Among these, SVM had the best prediction effect with prediction accuracy reaching 0.9867. The results proved the feasibility of using a detection method based on multiple features image fusion and machine learning, providing a theoretical reference for the nondestructive detection of chicken breast freshness.

There has been a recent increase in the frequency of mass disaster events. Following these events, the rapid location of victims is paramount. Currently, the most reliable search method is scent detection dogs, which use their sense of smell to locate victims accurately and efficiently. Despite their efficacy, they have limited working times, can give false positive responses, and involve high costs. Therefore, alternative methods for detecting volatile compounds are needed, such as using electronic noses (e-noses). An e-nose named the ‘NOS.E’ was developed and has been used successfully to detect VOCs released from human remains in an open-air environment. However, the system’s full capabilities are currently unknown, and therefore, this work aimed to evaluate the NOS.E to determine the efficacy of detection and expected sensor response. This was achieved using analytical standards representative of known human ante-mortem and decomposition VOCs. Standards were air diluted in Tedlar gas sampling bags and sampled using the NOS.E. This study concluded that the e-nose could detect and differentiate a range of VOCs prevalent in ante-mortem and decomposition VOC profiles, with an average LOD of 7.9 ppm, across a range of different chemical classes. The NOS.E was then utilized in a simulated mass disaster scenario using donated human cadavers, where the system showed a significant difference between the known human donor and control samples from day 3 post-mortem. Overall, the NOS.E was advantageous: the system had low detection limits while offering portability, shorter sampling times, and lower costs than dogs and benchtop analytical instruments.

As the simplest feature extraction, traditional hand-crafted transient-state features have been widely used in the area of electronic noses (e-noses). However, the influence of noise in the calculation of numerical differentiation leads to inaccuracy and instability in extracting these features. To tackle this issue, a novel numerical differentiation algorithm is proposed, which uses kernel-based regularization. The proposed method can provide accurate and stable transient-state features by directly estimating high-order derivatives from the noise-contaminated sensor’s reading. The feature representation is a prerequisite for the good performance of e-noses. Nevertheless, it should be noted that this performance in real applications can still be affected by other factors, such as sensor drift and the disturbance of nontarget odors. These issues can be addressed by applying a framework of domain adaptation and one-class classification. The proposed method and the adopted framework are verified in a field experiment, which identifies the odor of four targets and two disturbance whiskies measured by a self-designed e-nose system. The classification accuracy with traditional features is improved from $\mathbf{71.90\%}$ to $\mathbf{86.36\%}$ , showing the good potential of the proposed method for application in the area of e-noses.

The occurrence of mass disasters are increasing as a result of changing climates and the growing threat of terrorist activities/conflicts. When these tragedies strike, it is critical to locate victims. While search and rescue dogs are trained to locate the living, cadaver detection dogs are trained to locate the deceased. These dogs rely on the volatile organic compounds (VOCs) emitted from the victims to do so. Knowing which dog unit to deploy can be challenging, and the victims’ makeup is likely to change following disasters in densely inhabited places, where commingling is likely to occur. The use of electronic nose technologies in forensic science is a recent breakthrough. Due to their ability to detect differing VOCs, this technology can be used to assist in the recovery of victims in disaster events. The most popular types of accessible gas sensor technologies are briefly introduced and compared in this article for their potential use to locate missing persons, both living and deceased. The current and future market needs are articulated, and the lack of enrichment of these needs is examined in relation to the capabilities of existing gas sensors. This will inform further research areas of preference to increase victim detection capabilities.

Brain-Computer Interfaces (BCIs) rely on accurately decoding electroencephalography (EEG) motor imagery (MI) signals for effective device control. Graph Neural Networks (GNNs) outperform Convolutional Neural Networks (CNNs) in this regard, by leveraging the spatial relationships between EEG electrodes through adjacency matrices. The EEG_GLT-Net framework, featuring the state-of-the-art EEG_GLT adjacency matrix method, has notably enhanced EEG MI signal classification, evidenced by an average accuracy of 83.95% across 20 subjects on the PhysioNet dataset. This significantly exceeds the 76.10% accuracy rate achieved using the Pearson Correlation Coefficient (PCC) method within the same framework. In this research, we advance the field by applying a Reinforcement Learning (RL) approach to the classification of EEG MI signals. Our innovative method empowers the RL agent, enabling not only the classification of EEG MI data points with higher accuracy, but effective identification of EEG MI data points that are less distinct. We present the EEG_RL-Net, an enhancement of the EEG_GLT-Net framework, which incorporates the trained EEG GCN Block from EEG_GLT-Net at an adjacency matrix density of 13.39% alongside the RL-centric Dueling Deep Q Network (Dueling DQN) block. The EEG_RL-Net model showcases exceptional classification performance, achieving an unprecedented average accuracy of 96.40% across 20 subjects within 25 milliseconds. This model illustrates the transformative effect of the RL in EEG MI time point classification.

The calibration and its associated experimental design scheme for the tri-axial magnetometers are explored in this paper. For the widely used 9-parameter model of tri-axial Magnetometers, based on experimental design of Tri-axial accelerometers, a 12-observation experiment scheme, whose observations are located in the vertices of an icosahedron, is presented to reduce estimation error. As this 12-observation scheme is proved to be rotatable, before experiment, it is not required to identify the direction of the magnetic field of the Earth. For this 12-observation experiment scheme, a simple parameter estimation algorithm is presented, which can be easily implemented in a micro-controller with low computational capacity. Although, this calibration method utilizes the projections of the local Earth magnetic field as calibration inputs, a high precision turntable is not essential for ensuring desired calibration accuracy.

Brain-Computer Interfaces (BCIs) rely on accurately decoding electroencephalography (EEG) motor imagery (MI) signals for effective device control. Graph Neural Networks (GNNs) outperform Convolutional Neural Networks (CNNs) in this regard, by leveraging the spatial relationships between EEG electrodes through adjacency matrices. The EEG GLT-Net framework, featuring the state-of-the-art EEG GLT adjacency matrix method, has notably enhanced EEG MI signal classification, evidenced by an average accuracy of 83.95% across 20 subjects on the PhysioNet dataset. This significantly exceeds the 76.10% accuracy rate achieved using the Pearson Correlation Coefficient (Pearson) method within the same framework. In this research, we advance the field by applying a Reinforcement Learning (RL) approach to the classification of EEG MI signals. Our innovative method empowers the RL agent, enabling not only the classification of EEG MI data points with higher accuracy, but effective identification of EEGMI data points that are less distinct. We present the EEG_RL-Net, an enhancement of the EEG_GLT-Net framework. The EEG_RL-Net model showcases exceptional classification performance, achieving an unprecedented average accuracy of 95.36% across 20 subjects within 18.20 milliseconds. This model illustrates the transformative effect of the RL in EEG MI real time point classification.

This paper presents an efficient in-field calibration method tailored for low-cost triaxial MEMS gyroscopes often used in healthcare applications. Traditional calibration techniques are challenging to implement in clinical settings due to the unavailability of high-precision equipment. Unlike the auto-calibration approaches used for triaxial MEMS accelerometers, which rely on local gravity, gyroscopes lack a reliable reference since the Earth’s self-rotation speed is insufficient for accurate calibration. To address this limitation, we propose a novel method that uses manual rotation of the MEMS gyroscope to a specific angle (360°) as the calibration reference. This approach iteratively estimates the sensor’s attitude without requiring any external equipment. Numerical simulations and empirical tests validate that the calibration error is low and that parameter estimation is unbiased. The method can be implemented in real-time on a low-energy microcontroller and completed in under 30 seconds. Comparative results demonstrate that the proposed technique outperforms existing state-of-the-art methods, achieving scale factor and bias errors of less than 2.5×10−2 for LSM9DS1 and less than 1×10−2 for ICM20948.

In this paper, a linear triaxial magnetometer calibration model with six observations is proposed for wearable Inertial Measurement Unit sensors. First, the information matrix of the six-parameter calibration model is derived under the proposed six-observation scheme, and its G-optimality is proven. Thereafter, a series of simulation studies were conducted to demonstrate the effectiveness and robustness of the six-observation scheme. Finally, the designed six-observation scheme is experimentally verified by the commercial Inertial Measurement Unit devices. The simulation and experimental results both demonstrate the efficiency and accuracy of the proposed optimal six-observation scheme. This scheme simplifies the calibration procedure and reduces the calculation workload, further providing guidance for the daily use of wearable sensor calibration.

This paper studies the problem of distributed integral controllability for non-square multiple-input-output (MIMO) processes, where the number of inputs is greater than the number of outputs. Distributed or decentralized integral controllability (DIC) is a property that allows the use of stable decentralized controllers combined with integral action to achieve offset-free control and independent tuning of any subset of the control loops. Previous research has focused on DIC for linear/nonlinear square processes and nonlinear multiple-input single-output (MISO) systems, but not for non-square MIMO processes. We extend the concept of DIC to non-square MIMO processes and derive sufficient and necessary conditions for distributed integral controllability, which is a generalization of DIC for square systems. We also present a simulation study to demonstrate the effectiveness of our method.